Highland, Utah — May 18, 2026

Picture a medical device sales rep on their first solo call. They're standing outside an operating room, about to pitch a surgeon on a new surgical implant worth a six-figure contract. They've read the product literature. They've sat through the onboarding sessions. But they've never actually practiced the conversation — not with a real objection, not under real pressure, not against someone who knows the clinical environment as well as the buyer on the other side of that door.

This is the moment SoloFire is trying to fix.

"A sales rep in medical device takes nine to twelve months to ramp," says Ben Mosbarger, who acquired SoloFire roughly 18 months ago. "It takes a long time to learn these products. A knee replacement, a catheter, a surgical implant — these aren't things you learn overnight. And until now, the only way to get better was to go on real calls with a sales manager riding along and hope you didn't burn a million-dollar lead in the process."

This month, SoloFire launched what Mosbarger said is a first-of-its-kind answer to that problem: an AI coaching product that lets medical device sales reps practice realistic sales conversations with AI-simulated buyers before they ever walk into a hospital. To Mosbarger's knowledge, no competitor in the medical device vertical has shipped anything like it.

From Content Tool to Full Platform

When TechBuzz last covered SoloFire following Mosbarger's acquisition, the platform was primarily a content management and version control solution for field reps in medical device, biotech, and life sciences. The core value was straightforward but important in a heavily regulated industry: make sure every rep in the field has the current, FDA-compliant version of every product document, clinical brief, and sales asset at their fingertips.

That foundation hasn't gone away. But over the past year, Mosbarger and his team have built significantly on top of it, transforming SoloFire into what he describes as a consolidated sales enablement platform, combining content management, SCORM-based LMS training, an AI sales assistant, and now AI coaching, all under a single contract.

"There are multiple systems that companies buy today," Mosbarger shared with TechBuzz. "They buy a learning management system, a content management system, a coaching system — all separately, all to train and enable reps and reduce the sales cycle. We've put all of that in one platform, with one contract and one point of contact. That's it."

The competitive targets are explicit. Showpad, Seismic, and Highspot dominate the broader sales enablement market, but none of them have gone deep into the specific workflows, regulatory requirements, and buyer dynamics of medical device sales. Mosbarger isn't trying to beat them at their own game. He's carving out the vertical they haven't fully claimed.

"Our competitors are Showpad and Seismic," he says. "Please go compare us to them. I would welcome that."

Why AI Actually Works Here

Before getting to the coaching product itself, it's worth understanding why SoloFire's AI implementation works when so many others don't. Mosbarger has a clear answer to that question.

"AI needs context," he says. "Without context, AI doesn't really help you that much. Everyone is saying AI this and AI that, spending a bunch of money on it, and hoping they find value. But our application is chock-full of content that is all context."

The insight is simple but significant. Every sales deck, product specification, clinical white paper, and marketing asset that a SoloFire customer loads into the platform becomes the AI's knowledge base. The system doesn't need to be separately configured or trained. It already knows the product, the positioning, the target buyers, and the clinical use cases — because all of that information already lives in the platform.

"If I've got all your sales material and all your marketing material, I know what your product is, I know how to position it, and I know to whom," Mosbarger says. "And so with zero training — literally zero — I can open the app and have it sell the product. It just knows."

He demonstrated exactly that on the day of our interview. A new biotech dental client had come onboard that afternoon. Their content was uploaded at 2 p.m. Within hours, Mosbarger was on a live call with the company's CEO, walking him through the platform. He had the CEO share his screen and asked him to let the AI sell him his own product.

It did. Unprompted and without any configuration, the AI walked through the dental product's clinical advantages, explained why grade-five titanium matters in oral surgery applications, and handled the conversation the way a trained rep would. The CEO had never seen it before. Neither had Mosbarger, in that specific context.

"It just knew it," Mosbarger says, "because all of his sales and marketing content knew it."

What AI Coaching Actually Looks Like

The AI coaching product — launched in this month — works by letting sales reps simulate real buyer conversations before they're in front of real buyers. A rep opens the app, selects a buyer persona: a VP of Surgery, a hospital procurement officer, a distributor, a frontline OR manager. They start a conversation. They can play the seller or flip the dynamic and watch the AI demonstrate how to handle it.

Because the AI is working from the company's own product content, it doesn't just produce generic sales coaching. It handles actual objections about actual products, responds to clinical questions with product-specific answers, and behaves like a buyer who knows the space.

After each session, the rep receives a scorecard. Their manager receives it too — giving sales leadership visibility into where new hires are developing and where they still have gaps, without ever having to be in the room.

"Instead of spending time with a sales manager sitting in the car, going on calls, burning real leads with a rep who isn't ready, now they can cycle through a hundred practice conversations with zero risk," Mosbarger says. "That's a big deal when you're talking about multimillion-dollar contracts on the line."

The AI coaching launch follows the January release of Spark AI, SoloFire's AI sales assistant. Where coaching prepares reps before a meeting, Spark AI supports them in real time. A rep can ask it natural-language questions: "What size needle do I use for this procedure?" or "I'm meeting with a VP of Surgery at Johns Hopkins — what in my bag should I lead with?" The rep then receives sourced answers drawn from their company's content library. The assistant cites its sources, so reps know exactly which document the answer came from.

Early Signals

Both features are new enough that Mosbarger doesn't yet have a roster of customers who switched from Showpad or Seismic to cite. But he says the market response has been immediate and qualitatively different from anything he's seen in his time with SoloFire.

"It used to take months to sell our product," Mosbarger said. "Now I've seen leads come in and say, let's talk in two days, let's talk in a week. You can feel it. The platform went from a nice-to-have content management system to a full solution that fills gaps they really needed covered."

Existing customers have strong incentives to upgrade. New prospects are encountering a platform that addresses training, coaching, and content management in a single purchase decision. It is a consolidation play that resonates with sales operations teams managing multiple vendor contracts.

The company has added team members in finance, accounting, and onboarding since the acquisition, but has kept the engineering organization lean. Mosbarger credits AI-assisted development for the team's ability to move quickly — he says features that would have taken weeks or months to build can now be scoped and shipped in days.

Pricing has adjusted since the acquisition as well, settling at approximately $35 per user per month, down from the $50 figure cited in TechBuzz's original coverage.

A Small Team, A Specific Bet

SoloFire is a small company making a focused wager: that going deep in one vertical, building the context and workflows that horizontal platforms don't prioritize, is more defensible than competing for the general market. The AI coaching launch is the clearest expression of that bet so far.

"Anything that enables a sales rep to reduce that ramp from years to months is going to make a real revenue impact for our customers," Mosbarger stated. "If we can do that, we're not just a nice software tool. We're a need-to-have."

For a platform that started as a way to make sure field reps had the right PDF at the right time, that's a significant evolution. Whether the market validates it will become clearer over the coming months as AI coaching moves from launch announcement to customer results.

Mosbarger, for his part, isn't waiting around to find out.

SoloFire is based in Highland, Utah. Learn more at solofire.com.

On the evening of May 14th, in a building that has witnessed more than 130 years of Utah history, a group of innovators gathered to imagine the state's creative future.

The Salt Lake Art Museum (SLAM), still in its pre-opening phase, occupies the historic B'nai Israel Temple at 249 South 400 East in Salt Lake City — a striking Romanesque Revival structure completed in 1891, built by Utah's early Jewish pioneer community. Its rusticated sandstone facade and distinctive dome have anchored that corner of the city for generations. Now the building is becoming something new: a home for Utah art, past, present, and future.

That evening, the museum's walls were still bare white, unfinished, waiting. So organizers did something inspired: they invited early supporters to draw on them. Silhouettes were traced, messages written, names left behind — a human layer that will eventually be painted over and invisible, but permanent nonetheless.

"When we paint over it," said SLAM's director, Dr. Micah Christensen, "we'll be able to tell everyone there's a drawing room underneath the Salt Lake Art Museum. You're a masterpiece — you just can't see it underneath there."

It was an apt metaphor for what X5 aims to build across Utah.

X5 — a convergence platform anchored symbolically in Utah's five national parks and designed to connect creators, technologists, entrepreneurs, and capital — chose that evening to lay out its vision to early supporters. The Salt Lake Art Show, running May 14–17 at the Mountain America Expo Center in Sandy, was X5's first public activation, and the leaders who would shepherd it gathered at the museum beforehand to explain what they were really after.

Joe Ross, Vice Chair and Chief Innovation Officer, was direct about the stakes. Sundance, he noted, had historically generated over $130 million in annual economic impact for Utah, with the majority coming from out-of-state visitors. With Sundance's departure, that impact needed a successor — and X5 was designed to be something larger and more durable than a film festival replacement.

"This is not a replacement festival," Ross said. "We're creating a multi-sector economic platform, something bigger, longer lasting, and structurally more powerful. We want to build something neutral and agnostic that works for everybody. The goal isn't to replace what was lost. It's to build something that exceeds it."

At its core, Ross explained, X5 is organized around three clusters: an industry and STEM cluster covering AI, energy, aerospace, and life sciences; a culture cluster encompassing film, music, art, gaming, and immersive media; and a capital and workforce cluster connecting founders and creators with investment, policy, and economic development. The target, over time, is 75,000 attendees, 300 conference sessions, and programming spread across five Utah cities from Provo to Ogden — with everything streamed to a global audience.

Kevin O'Keefe, President of Platform Operations, X5, Founder and CEO of Peaks Art Fairs LLC and the organizing force behind the Salt Lake Art Show, had spent the days before the event staring at thousands of square feet of bare white exhibition walls. He spoke about the ambition in practical terms: reaching an audience of 40 million people through existing media channels, drawing private equity and venture capital toward Utah creators, building a platform that makes Salt Lake City a destination the world returns to year after year.

"We want people to know they have to come to Salt Lake," O'Keefe said. "Not just to see what's happening, but because what's happening here is the future. When innovators, creators, and capital all come together in the same place, that's when things actually move."

SLAM's director is Dr. Micah Christensen. Notably, he's also a partner at Anthony's Fine Art & Antiques, a 30,000-square-foot Salt Lake institution that O'Keefe described as arguably the largest secondary art gallery in the western United States, if not the country. "You have to go look at that place," he urged the audience with enthusiasm. "On its own, it's its own museum."

David Steenhoek, Vice Chair and Chief Creative Officer, closed the evening on a more philosophical note, and a personal one. Steenhoek, who has deep Utah roots, spoke about what X5 is ultimately for.

"Now is the new renaissance," he said. "And the renaissance is us: creators and founders who have the support, the inclusion, the network, and the access to finance and innovation that they've never had before. Our society has become too focused on ego and competition. X5 is about the golden rule, about building something together. The ocean thrives because the coral and the fish are alive and healthy. We've been doing this wrong. It's time to work together."

He ended with a challenge: "Ninety-nine percent of people don't follow through or follow up. If you are just the one percent who does — it changes everything."

From West Valley to Amsterdam: Erik Jensen

Not every artist at the Salt Lake Art Show traveled far to get there. Erik Jensen grew up in West Valley City — still living, he noted with a laugh, in the home his grandfather built. Today he travels the world selling fine art. The bridge between those two facts is a pile of old computer keyboards.

Jensen is the founder of Erik Jensen Art, and his work is made entirely from recycled keyboard keys. Not decorative keyboard-themed art, but actual keys, sourced in bulk, sorted by color, and painstakingly arranged into portraits and compositions that reward a close look. Hidden within many pieces are words and messages, legible only to viewers who lean in.

"They're all recycled keyboard keys," Jensen explained, standing next to one of his pieces at the show. "Mostly from the '90s. And there are hidden messages, words worked into the art that most people don't notice at first."

The color palette is where Jensen's process gets particularly inventive. Black keys are plentiful, he said, but the beige, white, and gray keys from older machines are what he's always hunting for. The tan keys of the 1990s — once aesthetically unappealing — become highly desirable raw material for his color work. Jensen spent roughly a year researching plastics and chemistry before developing a soaking-and-dyeing process that transforms those drab keys into rich, varied hues. The longer they soak, the darker they get; pull them early for lighter tones.

"That's my trade secret — the coloring," he said with a grin.

The idea came from a college assignment: take something people no longer want and make it into something again. Jensen grabbed a discarded keyboard and fashioned a small face. His professor's reaction — "that's so cool, you have to do more" — sent him down a research rabbit hole. He discovered that practically nobody was recycling keyboards, and certainly nobody was making fine art from them. He saw the opening and took it.

That was about eleven years ago. Jensen spent five years as a high school art teacher before leaving to pursue his work full time eight years ago. The gamble paid off. He has since converted 1.4 million individual keys into artwork, sold pieces to collectors in 35 U.S. states and 16 countries, and counts the tech industry among his biggest buyers — fitting, perhaps, for art made from the tools of the trade.

"Most of my stuff ends up in tech," he said.

Just before the Salt Lake show, Jensen returned from Amsterdam, where he completed a commissioned piece for Logitech at their European offices. He was heading to New York for another major show the following Tuesday.

Not bad for a kid from West Valley.

Where Every Bead Is a Mountain: Mosgaadace Casuse

Not every artist at the Salt Lake Art Show brought work that could be called sacred. Mosgaadace Casuse did.

Casuse, who is of Navajo (Dene) and Ojibwe (Anishinaabe) heritage, grew up in Santa Fe and now splits his time between there and Phoenix. He has been making art in the most literal sense since he was five years old, when his mother and father first sat him down to learn lapidary and jewelry-making. Two small pieces on his table at the show were from that period. Made using tufa casting with sterling silver, they were among the last two works he still has from childhood.

"My mom and my dad taught me jewelry when I was five years old," he said. "I started with lapidary and painting."

Tufa casting is a technique primarily practiced by Navajo artists. The artist carves a design into soft tufa stone — a porous volcanic rock — then pours molten silver directly into the mold. The result carries a distinctive organic texture impossible to replicate by other means. Casuse's work is represented at Malouf on the Plaza, with locations in Santa Fe and New York, as well as through Four Winds Gallery on consignment. His Instagram handle is @mastermukwa.

But the piece that stopped visitors in their tracks was one he did not make alone.

Casuse described a beadwork and silversmithing piece created together with his mother and father as a family work. It carries an oral story: long ago, a turtle came and took the songs from the Native people. As he walked away into the mountains, he left them with a message — when you learn silence, I will return song. The piece traces that journey. Each bead represents one of the mountains of North America, moving southward until arriving at the Four Corners region of the Navajo nation. Hidden inside, worked into a silver band, are those words: when you learn silence, I will return song.

"As our people travel through each mountain, trying to find the turtle and find our songs, we learn to be silent with the land," Casuse said.

Another piece carried a different ceremony: the Night's Way, observed in early winter for elders. Casuse rendered it in a progression of color: blue sky giving way to coral, then to the deepening dark of night, anchored finally by a black star sapphire. A week after Malouf on the Plaza purchased that piece, a visitor from London acquired it and brought it to a museum there.

Casuse also showed bowl ties incorporating something unexpected: circuit boards. That innovation traces directly to his mother, who is Anishinaabe from the Great Lakes region of Minnesota. The Anishinaabe people traditionally work with birch bark, using it for canoes, dwellings, and intricate artwork. His mother was a pioneer in translating those birch bark techniques into metal smithing. Casuse adapted the same methods to incorporate circuit boards, creating a conversation between ancient material culture and the present.

"She was the first person to innovate that and bring it into metal smithing," he said. "I use those techniques for circuit boards."

Find more of Casuse's art at instagram.com/mastermukwa/.

Steel, Fire, and a Woolly Mammoth: Clinton Lesh

Clinton Lesh makes art built to survive a crowd. Literally.

The Bozeman, Montana sculptor brought a life-sized stainless steel woolly mammoth to the Salt Lake Art Show. It is the same piece he debuted at Burning Man last year, where festival-goers climbed on it, sat on it, and posed with it across the alkaline flats of the Nevada desert. It held up fine.

"We got some good pictures of people sitting on it," Lesh said with a laugh.

The mammoth was a Burning Man grant project, conceived around last year's festival theme of futuristic technology. Where other artists went predictably forward — robots, spaceships, chrome abstractions — Lesh went the other direction, into deep time.

"I looked into the past, at dinosaurs and ice age animals," he explained. "And I found out they are cloning the mammoths. So they're a futuristic animal; they're just not here right now."

It's a sharp conceptual hook. Colossal Biosciences and other biotech firms are working to resurrect the woolly mammoth by splicing its genes into Asian elephant embryos. Lesh saw that and recognized a sculpture waiting to be made.

The execution is as impressive as the concept. The piece is fabricated entirely from stainless steel, then heat-treated to produce its warm, otherworldly coloring. Stainless steel begins turning gold at around 600 degrees, Lesh explained. Keep heating, and it shifts through amber into blue and purple. The result is a surface that looks almost organic, glowing at dusk in a way that plain steel never could. At Burning Man, on the Playa at golden hour, it stopped people in their tracks.

The eyes are hand-blown glass, made by an artisan in Montana, chosen because Lesh wanted them slightly imperfect — realistic rather than pristine.

The mammoth is not his only large-scale festival piece. The year before, he brought a giant wolf sculpture to Burning Man. He applied for a grant this year with a proposed 15-foot Pegasus. His work has also traveled to the Loveland Sculpture in the Park show in Colorado, billed as the largest outdoor sculpture show in America.

Back home in Bozeman his sculptures live outside his studio, visible from the road. More of his work can be found at clintonleshsculptures.com and at www.instagram.com/leshclinton/

A Vision Bigger Than the Show

Walking the floor of the Salt Lake Art Show, past keyboard-key portraits, beaded oral histories, heat-patinated steel, it was easy to see what X5's founders are reaching for. The artists who came to Sandy that weekend were not an afterthought to the platform's technology and capital ambitions. They were, as Steenhoek put it, the foundation.

The building that hosted the X5 launch party the night before the show opened tells a version of the same story. The B'nai Israel Temple was built in 1891 by early Utahns who wanted to create something lasting, a community institution that would outlive its founders, adapt to new purposes, and remain standing long after the circumstances of its creation had changed. It has done exactly that.

X5 is making a similar bet. The silhouettes drawn on those museum walls that Thursday evening will eventually be painted over, invisible beneath whatever the Salt Lake Art Museum becomes. But they will still be there, an early layer, a founding mark, a reminder that something began here.

"We don't want to create unicorns," Steenhoek said that night. "We want to create supernovas."

The art show that followed suggested they might be onto something.

Healthcare and technology leaders are partnering with Utah Valley University to launch two new competitions aimed at accelerating innovation in artificial intelligence, digital health, and digital identity systems.

The programs, organized through a partnership between HITLAB (Healthcare Innovation and Technology Lab) and Utah Valley University, are designed to connect students, startups, researchers, and entrepreneurs with investors, mentors, and healthcare industry leaders.

Healthcare innovation often struggles to move from research into real-world patient care because of limited funding, weak industry partnerships, and lack of institutional support.

“The gap between breakthroughs and patient care is still too wide, and it costs lives,” said Stan Kachnowski, Chair of HITLAB. “This competition closes that gap by connecting innovators directly with the mentorship, capital, and industry relationships that turn promising solutions into deployed reality.”

HITLAB is a New York City-based digital health research organization focused on digital health research, evaluation, and innovation. It uses evidence-based methods to test and validate health technology before bringing it to market. One of its published studies produced an algorithm that identified infant babbling patterns linked to early developmental issues with 91% accuracy.

The initiative will feature two separate programs running simultaneously: a university hackathon and the Healthcare Innovation World Cup.

The University Hackathon

Tyler Small, Senior Director of UVU's Kahlert Applied AI Institute, said the event is intended to create meaningful opportunities for both students and businesses.

“This isn’t a fake-problem type of hackathon,” Small said. “Organizations are paying money to have real technical problems solved. Companies attending will be looking for interns, full-time employees, and technical talent. It creates a direct bridge between students and industry leaders.”

The hackathon will also feature a bounty system, in which companies present unsolved technical challenges and attach cash prizes to successful solutions. According to Tyler Jennings, three companies are currently signed up offer bounties in 2026: Philo Ventures, Waystar, and Halda.ai. Organizers say they expect that number to grow.

Companies provide detailed descriptions of their challenges, and participating teams spend the 24-hour competition developing solutions before presenting them to judges and company representatives. Winning teams receive the funding attached to each bounty.

“We at the Utah Governor’s Office of Economic Opportunity and The Nucleus Institute were thrilled with what teams built for our projects, and we’re excited to work with our bounty winners to bring these projects to life,” Jennings said.

Jennings also works as the local managing director for Sandbox, an entrepreneurial program focused on helping students launch businesses while earning academic credit.

“This partnership with HITLAB, the Governor’s Office, and the Utah System of Higher Education represents exactly what we’re building at UVU and the Smith College of Engineering and Technology,” said Barclay Burns.

“These collaborations create the ecosystem our students need by connecting academic innovation directly to industry networks, strategic mentorship, and the resources that turn classroom potential into real-world impact,” Burns said.

Other faculty members also expressed support for the event.

Majid Memari, who also serves as an NVIDIA ambassador and principal AI architect at the Gary R. Herbert Institute for Public Policy, said the competition offers students a more practical learning experience.

“I’m excited that we are hosting this hackathon,” Memari said. “This semester I took students to two conferences, but they wanted a more applied learning experience. This gives them an opportunity to build something meaningful.”

The Healthcare Innovation World Cup

The Healthcare Innovation World Cup is designed for startups, researchers, and innovators with validated concepts, prototypes, or early-stage companies.

Organizers say the competition is open to participants ranging from pre-seed startups to companies at the Series A stage. Academic research teams, graduate students, and interdisciplinary innovators with developed prototypes are also encouraged to apply.

Teams will compete in three innovation categories:

• Applied AI
• Healthcare and Life Sciences
• Digital Identity and Data

According to the HITLAB website, participants will receive mentorship from healthcare and AI leaders, pitch coaching, and opportunities to connect with investors, strategic partners, and pilot programs.

Burns said the event also reflects Utah’s growing role in digital identity development.

“The World Cup continues to position Utah and UVU as leaders in the mission-critical field of digital identity,” Burns said. “Secure digital identity systems are becoming foundational to healthcare, education, finance, life sciences, and emerging technology sectors.”

Organizers say they are seeking high-impact solutions that demonstrate strong problem-solution fit, technical feasibility, scalability, deployment readiness, and thoughtful consideration of ethics, privacy, and equity.

Applications for both competitions opened May 12 and close June 3. Finalists will be announced June 10, and the live final pitch competition and hackathon will take place June 17 and 18 at Utah Valley University.

More information is available at HITLAB World Cup.

Salt Lake City, Utah — May 15, 2026

The last major session of the 2026 American Causal Inference Conference was titled "Are We Being Replaced? Causality in the Age of AI." It ran Thursday morning in Salon E-F of the Salt Lake City Creek Marriott, and by the time it ended, the audience had heard three very different answers to that question from three of the field's leading researchers.

Then a UVU professor, Dr. Brian Knaeble, held a meeting for a group of Utahns to figure out what the conference meant for them.

Knaeble, UVU professor of computer science and president-elect of the Society for Causal Inference (SCI), opened a free networking session he had organized for Utah's broader tech and data science community. Graduate students sat next to working engineers. A health IT specialist described a feedback loop problem in EMS care. A recent data science grad wondered aloud how to bring what he'd just learned back to his company. A UVU undergraduate asked where to even begin. A BYU team showed up to weigh in and share their perspectives.

It was a sincere conversation: people without tenure or speaking slots trying to figure out what four days of doctoral-level mathematics actually meant for them.

What had just happened in the final plenary session was worth understanding.

What Is Causal Inference

Causal inference, the focus of the conference, is the process of determining whether and to what extent an action causes a specific outcome. It goes beyond standard statistics and correlation, aiming to answer "what if" questions and isolate cause and effect in complex systems. The distinction matters most when organizations have to decide between options rather than simply predict an outcome.

"If you're wanting to make a decision on whether to do X or not, causal inference can help," one meetup attendee told the group. "A lot of machine learning has become very, very good at forecasting and predicting. When the question turns into a 'what if' question — like 'what would happen if we double the cost of our product?' — that question becomes a causal inference problem."

The field is growing fast. Knaeble said UVU's course enrollment tells the story plainly: "UVU had zero causal inference students a few years ago. Now we've got three or four sections of the course. It just keeps exploding."

Three Researchers, Three Answers

Moderator Rohit Bhattacharya of Williams College, a former PhD student of one of the panelists, now an assistant professor, opened the session with a challenge. Before worrying about whether AI will replace us, he said, it's worth asking what we even mean when we say AI "solved" a problem. He described two researchers who had recently used LLMs in wildly different ways: one to prove a mathematical lower bound, another simply to get a book recommendation that led to a solution. Both called it AI assistance. Neither experience was the same.

He also noted who, precisely, is most worried about replacement. Not the tenured faculty in the room. It's younger people — undergraduates staring down a job market being reshaped faster than any curriculum can adapt. He cited a commencement speaker at the University of Central Florida who called AI the defining tech revolution of the generation. The crowd booed. When she referenced a time before AI, they applauded.

That detail set the emotional temperature for everything that followed.

Emre Kiciman: More Problems, Not Fewer

Emre Kiciman, Partner Research Manager at Microsoft, opened with an argument that was deliberately optimistic. AI isn't narrowing the scope of causal work, he said. It's expanding it into territory the field has never reached before.

His central insight: enormous amounts of human reasoning are implicitly causal but have never been treated that way. Business strategy documents, policy impact assessments, legal briefs, financial analyses — all of these make causal claims. They assert that X caused Y, that a policy change will produce a specific outcome, that a trend will continue because of identifiable forces. And all of them make the same kinds of errors that formal causal analysis is designed to catch: missing confounders, implicit assumptions left unexamined, competing hypotheses evaluated by instinct rather than rigorously.

Kiciman's team has been developing tools that apply causal methods to these narrative documents. The process extracts an implied causal graph from the text, critiques it for structural problems, and then runs simulations — using LLMs to fill in values — to test how sensitive the document's central claims are to what it left out.

He demonstrated the approach using a financial podcast in which an analyst argued that AI infrastructure investment is fundamentally demand-driven and sustainable. When his team pulled in outside sources, they found significant pathways the analyst had ignored — financial fragility from bank lending, rapid hardware obsolescence, structural risks to the investment thesis — that measurably shifted the range of likely outcomes.

"I'm not worried about AI replacing this anytime soon," Kiciman said. "I think it's more expanding our scope and the places where the work we do can be applied."

Kun Zhang: AI Needs Causality to Grow Up

Kun Zhang, who holds appointments at Carnegie Mellon University and the Mohamed Bin Zayed University of Artificial Intelligence, agreed that causal methods can strengthen AI. However, his argument ran just as forcefully in the other direction. AI, as currently built, has deep structural problems that only causal thinking can fix.

He opened with a demonstration that landed immediately. Ask a leading image-generation model to produce "a peacock eating ice cream," and you get something that fails in revealing ways: beautiful peacocks with no ice cream, or the two elements awkwardly forced together. The models know peacocks. They know ice cream. What they cannot reliably do is combine concepts they haven't seen combined, because they've learned to map inputs to outputs rather than to reason about how things relate at the level of underlying structure.

The controllability problem is related. Zhang showed a leading AI image tool being asked to add a mustache to a generated face: the mustache doesn't appear, or the entire face changes. Asked to make a smiling girl in a garden look surprised: the expression shifts but so does her hair, her clothing, the scene's lighting. These aren't aesthetic glitches. They reveal something structural. Current models cannot isolate an intervention on one variable while holding others fixed, because they have no explicit representation of which features are causally independent of which others.

Zhang's lab has developed approaches built on causal representation learning that address exactly this — decomposing images and text into structured, independent concepts and learning the relationships between them, rather than brute-force mapping prompt to pixel. The results, which he demonstrated side by side against leading commercial models, showed his method changing precisely what the user asked to change while leaving everything else intact.

The deeper point extended from images to society. Zhang presented a structured analysis of AI's likely impact over the next two decades, with hard numbers: 72% of enterprises currently have AI in production, but only 18% report real productivity gains. That 54-point gap between capability deployment and realized value, he argued, is the defining fact of the current moment.

His timeline projects three phases: sharp near-term declines in entry-level hiring and early signs of skill atrophy, followed by a difficult adaptation period in which deeper risks become visible, an expertise pipeline crisis, growing inequality between those who use AI to amplify genuine knowledge and those who simply depend on it. A new equilibrium arrives eventually, but only if we navigate the transition well. Historical technology shifts have typically required 15 to 25 years of adaptation. AI, Zhang warned, may compress that into 5 to 10.

His most pointed warning concerned human cognition. Long-term, he argued, society risks bifurcating: AI-complementary experts who use the technology to amplify deep knowledge, and AI-dependent workers who can no longer function without machine assistance.

"We need to come up with a way to force students and everyone to try to be independent users of AI," Zhang said, "instead of being slaves of AI."

Ilya Shpitser: The Warning Nobody Wanted to Hear

Johns Hopkins' Ilya Shpitser delivered the session's sharpest remarks, and the ones that resonated longest in the Solitude Room afterward. The title he had given his talk was "Living With Blockhead," a reference to a thought experiment in philosophy of mind: a "blockhead" is a hypothetical system that passes every behavioral test for intelligence without any genuine understanding behind it. As a title for what became a pointed argument about AI and institutional power, it was either very dry or very apt.

He began on technical ground. Hallucinations — plausible but false outputs from LLMs — are not, in his view, a fixable bug. They are a fundamental property of systems built to predict. Predictors produce wrong outputs. Identifying those wrong outputs requires expertise. People who lack that expertise are at serious risk of a false sense of security, and the consequences scale with the stakes of the domain. Some tasks, he argued, are simply inappropriate for LLMs, not because the models are immature, but because of what they are. NP-complete problems won't yield to statistical pattern matching. High-stakes decisions where an incorrect output is catastrophic create accountability problems that don't disappear just because the model improves.

But Shpitser's most pointed argument was not technical. It was historical.

"I don't think we're being replaced," he said. "I think we're potentially being subordinated."

His claim: the real threat of AI in academia is not that it will do our jobs. It is that it will become one more instrument in a longer trend, the corporatization of the university, that weakens the independence on which scientific credibility depends. Pressure to adopt AI tools is arriving simultaneously with pressure to adopt corporate governance models, to prioritize donors and industry partnerships, to operate on administrative timelines that have nothing to do with how knowledge is produced.

He traced the stakes of that independence through a personal story that stopped the room. He had recently been at Cambridge for a causal inference event. One evening he attended evensong, the traditional choral service held in the college chapels. As the clergy processed in, he noticed their robes, and then had a moment of recognition. Those weren't clergy robes that looked like academic robes. They were the same robes. The hood, the cut, the whole garment, are identical to what academics wear at graduation ceremonies and PhD defenses. They are the same robes, inherited directly.

The modern university, Shpitser argued, did not emerge from nowhere. It grew out of medieval cathedral and monastic schools, institutions that derived their authority precisely from being independent of the two great centers of power in medieval Europe: the nobility and the powerful merchant families. The scholarly guilds that eventually organized themselves into universities carried that independence forward as a structural feature, not an accident. It was the condition of possibility for everything the academic project has since accomplished.

The robes are a remnant of that history. So, he argued, are peer review, tenure, and the norm of publishing results regardless of who funded the research. These aren't merely traditions. They are the institutional skeleton of independence, built over centuries to keep inquiry free from the distorting pressure of whoever holds power at a given moment.

"At the end of these trends," Shpitser said, "is the end of science as an independent human enterprise."

His prescription was blunt: faculty need to organize politically, not just academically. Coordinate across institutions. Push back against corporate capture. He acknowledged that STEM faculty are not historically good at this. His view is that this has to change.

The Room Pushed Back

The Q&A was among the liveliest of the conference.

One audience member pointed out that LLMs are themselves causal systems. Their training is a causal process. Their outputs can be intervened upon. Their internal representations can be studied with exactly the tools this community has developed. Was that an opportunity rather than a threat? All three panelists agreed it was, though Shpitser observed that the causal structure inside a large language model and the causal structure of the natural language it was trained on are two different things, and confusing them could be disastrous.

Another audience member challenged Shpitser directly: humans make mistakes too, often for the same reasons as LLMs, such as insufficient information, cognitive overload, systemic bias. At least with a model, you can measure the error rate. Shpitser's response was pointed: you can sue a doctor for malpractice. Accountability structures exist for human decision-makers in ways they do not yet exist for AI systems, and that asymmetry matters enormously in domains where errors cause real harm.

A diverse group of locals meets after the conference

Then the session ended, and a smaller room tried to figure out what to do next.

The Solitude Room gathering reflected how broadly this field's implications now reach. UVU undergraduates finishing degrees in software engineering and data science sat alongside a Westminster professor, an NYU professor of data science originally from Salt Lake City, BYU and University of Utah faculty, a Chief AI Officer, and working industry professionals. Someone mentioned the Great Salt Lake. Someone mentioned the 2034 Olympics. Someone mentioned manufacturing.

Dr. Knaeble is supportive of mutually beneficial partnerships between Utah's universities and its industrial sector. While artificial general intelligence requires causal reasoning, many machine learning practitioners in Utah are unaware of causal inference as a discipline. That gap, in his view, is an opportunity.

Barclay Burns, UVU's Chief AI Officer and Assistant Dean of the Smith College of Engineering and Technology, framed the stakes in foundational terms.

"We need to move much more to the core model — building, understanding the behavior of the models, and improving the models," Burns said. "If you can utilize math and causal inference to improve the behavior of the models themselves, particularly the behaviors within agents and between agents — that would position Utah as a foundational contributor."

Jingpeng (JP) Tang, Master of Computer Science Program Director at UVU, raised the possibility of pairing master's capstone projects with industry partners on causal inference problems — a way to close the loop between the conference's ideas and Utah's actual industry needs.

What This Can Look Like

Attendees offered concrete examples of where causal methods could matter right now.

David Balcombe, a UVU undergraduate and University of Utah graduate working in health IT, described a prototype he is building for emergency medical services that uses causal AI to address a problem he encounters constantly. Paramedics arrive on scene, form a primary and secondary diagnosis, administer treatment, and transport the patient to the emergency room. Once they hand off the patient, their involvement ends. And they almost never learn whether their field impression matched the hospital's final diagnosis.

"EMS doesn't have a feedback loop," Balcombe said. "They're clueless about anything that happens in the hospital."

His prototype would close that loop, eventually allowing agencies like Unified Fire to run their own quality improvement programs and compare patient outcomes across different equipment used in the field. It is precisely the kind of application Knaeble has in mind: a real industry problem, a genuine causal question, and a solution that requires more than prediction.

The discussion also surfaced harder questions. One participant, drawing on experience at a Stanford event connecting academics with Bay Area companies, raised the issue of compensation: if industry profits substantially from academic knowledge, what's the fair mechanism for sharing that value? No one had a clean answer. But the question itself suggested how seriously the group was thinking about building something sustainable

Several attendees discussed possible venues for a recurring meetup, interim spaces at various institutions, VC meeting rooms, co-working spaces such as Kiln, until ideally Convergence Hall, one of the planned buildings at The Point innovation campus on the former state prison site, opens in a few years.

Knaeble closed the meeting the same way he had opened it: with discipline about scope. "We should really settle on a niche and be really good at that small, narrow thing," he said.

Utah's Next Move

The contrast between the two rooms — the plenary and the Solitude Room — was instructive. In the Marriott's Salon E-F, three of the world's leading causal researchers debated whether AI will replace, subordinate, or simply redirect scientific work. In the Solitude Room, people who had just heard those arguments tried to translate them into something actionable.

Shpitser's warning carries particular weight in that context. The tools are genuinely powerful. The questions about who controls them, and to what ends, are genuinely open. Burns's call for Utah to become a foundational contributor — not just a consumer of AI tools built elsewhere — is one answer to those questions. Knaeble's instinct to pick a narrow problem and compound credibility slowly is another.

If causal AI is, as industry analysts project, on track for significant annual growth over the coming decade, Utah's proximity to this week's conference, and to its incoming Society for Causal Inference president, may matter more than the state currently realizes.

The answer to "are we being replaced?" may depend less on what the researchers decide than on whether the people in rooms like Thursday's Solitude Room, the ones willing to stay after the session ends and actually work on the problem, decide to keep showing up.

The American Causal Inference Conference was held May 11–14 at the Salt Lake Marriott Downtown at City Creek. The next conference will be held in Pittsburgh. The annual meeting of the international Society for Causal Inference draws researchers from around the world. Brian Knaeble can be reached at bknaeble@uvu.edu.

Lehi-based legal software company NetDocuments has introduced a redesigned version of its platform centered on what it calls a “legal context graph,” a system designed to connect documents, matters, communications, and institutional knowledge across a law firm’s records.

The company says the platform continuously maps relationships between legal documents and related activity while maintaining existing security permissions and ethical walls. A private preview opened this week.

The update reflects a broader shift in legal technology toward AI-assisted workflows. Gartner has named "context engineering" a strategic priority for AI leaders. Foundation Capital has called context graphs the next defining shift in enterprise AI, arguing the value lies not in who owns the data but in who can explain why decisions were made. Without that foundation, AI hallucinates, duplicates work, or misses the precedent that matters most. With it, AI becomes meaningfully more useful.

Rather than relying only on documents uploaded into a single AI session, NetDocuments says its system allows AI tools to work from a firm’s broader repository of institutional knowledge.

According to the company, lawyers opening a matter in the new interface can automatically see summaries, timelines, related precedents, key parties, and prior work connected to the matter. The platform also surfaces lawyers inside the firm who have handled similar work previously.

NetDocuments said the system was developed in partnership with Amazon Web Services and Elastic to process and organize large volumes of legal data.

The company is also positioning the platform as infrastructure for AI agents used inside and outside its software ecosystem, including integrations through its ndConnect framework with tools such as OpenAI’s ChatGPT and Anthropic’s Claude.

Among the features highlighted in the launch:

• Semantic search across firm documents and communications
• Automated matter summaries and timelines
• AI-assisted drafting and document comparison tools
• Permission-aware AI access to firm knowledge and precedent
• Integration with Microsoft 365 applications

CEO Josh Baxter said the company’s goal was to create a system capable of understanding relationships across legal data rather than simply storing files.

“Legal data is fundamentally different," he said. "It is language, not fields, and unlocking its meaning requires understanding it as a connected whole — every matter, every document, every communication, at firm scale. That kind of context has never been engineered into a legal platform before. It is what we have built, and it is the foundation lawyers and AI agents both need.”

The new platform will operate alongside NetDocuments’ existing interface, allowing firms to transition gradually rather than migrate all users at once.

NetDocuments, headquartered in Lehi, UT, with offices in Australia, the UK, Mexico and Brazil, is a software company producing cloud-based content and productivity tools for law firms, corporate legal teams, and compliance departments. Founded in 1999, NetDocuments' platform is used by over 7,000 law firms, corporate legal departments, and public sector organizations.

The company says that broader availability of the redesigned platform is expected in the coming months following the private preview phase. Demo videos covering the matter overview, contextual search, and agentic workflows, along with a deep dive on the architecture and the legal context behind it, are available here.  

Interested parties may join a waitlist or register for a June 9 webinar to see the new platform in action.

Joshua Ahlstrom had about $20 in his bank account when he started his first company. It was the spring of 2020, COVID had just shut down the country, and Ahlstrom — a door-to-door salesman recently out of a job — had a car cleaning kit his cousin had gifted him for Christmas and not much else. He started knocking on doors in his neighborhood, asking if he could detail people's cars for $20 a pop.

Five years later, Wild Detail has serviced roughly 4,000 vehicles. But it's the second company Ahlstrom built out of that detailing hustle — a small product startup called Tilted Spray Co. — that has become both his biggest success and his most frustrating fight.

The idea behind Tilted is almost embarrassingly simple. Anyone who has ever used a spray bottle knows the problem: tip it at an angle, and it stops spraying. There's not enough liquid to cover the dip tube intake. For a home cleaning enthusiast, that's a minor annoyance. For a professional detailing shop burning through a dozen trigger heads every two months, it's a persistent, expensive headache.

Ahlstrom's solution was an adapter, a small polypropylene piece that fits inside any standard spray bottle and uses a weighted, flexible silicone tube to follow the liquid no matter how the bottle is tilted. It sounds simple because it is. But finding a version that was affordable, durable, and universal took him three years of tinkering, a failed resin-printing operation, and a lucky connection to the founders of Utah belt-and-wallet maker Grip6, whose injection molding equipment made the final design possible.

"The versions that already existed cost about $20, and when the trigger head stopped working, the whole system needed to be replaced," Ahlstrom said. "We brought that down to two or three dollars. That was the idea: make 360-degree spray bottle accessibility five times cheaper."

From the Garage to a Million Views

By September 2024, Tilted had packaging, a website, and a product ready to ship. Ahlstrom and his business partner Cole Correa, who had merged his own detailing business with Wild Detail the previous spring, drove to Las Vegas for SEMA, one of the largest automotive trade shows in the world. They didn't have a booth. They just walked the floor, showed the adapter to anyone who would look, and collected feedback.

They forgot to film almost any of it.

The real launch happened the day after they got home, in Ahlstrom's garage. He made a simple video: a spray bottle before the adapter, struggling to spray on its side; a spray bottle after, working at any angle, getting every last drop. The video cracked a million views. Then another video did. Then another.

"It seemed like every video we put out was getting tens of thousands to even millions of views," he said.

Orders followed. Within a few months, Tilted had sold roughly 45,000 units and shipped more than 10,000 packs to customers around the world. Revenue surpassed, then doubled, what the detailing business was bringing in. Ahlstrom, who was running both companies while raising three kids, had built something real.

He just hadn't gotten around to listing it on Amazon yet.

Nine Pages of Fakes

Ahlstrom had a theory about Amazon: if he waited until the product's reviews and return rates were dialed in before listing it there, the early reviews would be strong enough to matter. It was a reasonable idea. It cost him.

One afternoon, sitting with his uncle Paul — a business connection who had warned him about manufacturers in China copying successful products and undercutting their originators — Ahlstrom decided to search Amazon for "360 spray bottle adapter," just to see what was out there.

"I go on there, and I see my face," he said. "A sponsored video of my face being used to sell a tilted adapter. With my product photos. My face, with their watermark on the video, where they used AI to change the background. It was my garage and my ceiling, but they had added a new couch and some plants."

He kept scrolling. He found nine pages of listings. Every one was slightly different: a product image copied and pasted three times, or an AI-distorted version of his own photography, or a lifestyle photo lifted directly from the Tilted website, sometimes showing one of his employees. In the descriptions, competitors had been careful: no mention of "Tilted Spray Co." by name. Just close enough to pass.

Entire Facebook pages had been built around his videos, with usernames like "josh.hedishin." a fake Chinese storefront that was racking up hundreds of thousands of views by reposting his content with altered backgrounds. In the comments, automated or human responders posed as Ahlstrom, answering customer questions. When someone asked whether the adapter worked with bleach, the fake "Josh" said yes. It doesn't.

"It kind of just stopped my heart a bit," Ahlstrom said.

The listings, he estimates, were diverting more than $10,000 in monthly sales, sales from customers who, having seen nearly 7 million views of Ahlstrom's face demonstrating the product, likely believed they were buying the real thing.

The Takedown Math Doesn't Work

Ahlstrom spent a weekend submitting 265 takedown notices to Amazon. Five were approved. To date he has submitted over 500 takedown notices to Amazon.

The ones that succeeded were straightforward: listings that used the trademarked names "Tilted" or "Tilted Spray Co." For the rest, Amazon's position was that the alterations to the images, a copied photo duplicated three times, or an AI remix of an original, were sufficient to make them distinct. No copyright violation, no takedown.

"You can take someone's product photo, copy and paste it three times, add a bottle from another lifestyle photo, and now it's its own image no longer covered under copyright," Ahlstrom said. "And even if the photo gets disabled, all they have to do is upload a new one. As long as it doesn't say 'Tilted Spray Co.' in the description, there's nothing we can do."

Professional takedown services quoted him $50 per listing. With hundreds of listings being added faster than they could be removed, the math was impossible. The time it took to successfully remove five listings, Ahlstrom said, more than a hundred new ones had appeared.

"The amount of effort it takes to take down a product," he explained, "is exponentially more effort than it is to list one. You'll just never catch up."

Ahlstrom believes the listings are being generated using AI agentic tools, automated systems that scrape product data, generate variations on images, and bulk-upload listings across platforms. It's not just Amazon. Fake storefronts selling counterfeit or nonexistent versions of Tilted have appeared on Alibaba, AliExpress, Temu, and across Meta's platforms. Ahlstrom purchased a couple of the Amazon listings. After several weeks they arrived.

"They weren’t even real copies," he said. "They didn't even match the photos that were on the listing. It was just a ploy to knock us down while they worked on it. And it shows in the Amazon reviews," he shared with TechBuzz.

The broader problem, Ahlstrom argues, is structural. Chinese counterfeiters operate largely outside the reach of U.S. patent, copyright, and trademark law. Legal action against individual sellers means, in practice, suing Amazon for allowing them to operate. It is a slow, expensive process with uncertain outcomes.

He referenced a YouTube creator named Vanader who faced 184 copycat listings of a patented product and spent 11 months and hundreds of thousands of dollars getting them removed — only to find the same product still available, this time with a different logo.

"I believe almost every single product market in the world has some form of this," Ahlstrom said. "China just doesn't follow U.S. patent, copyright, or trademark laws. And Amazon prioritizes customer experience, because happy customers spend more money. The problem is, that comes at the expense of the sellers."

What You Can Do (And What You Can't)

Ahlstrom is not without advice for founders in similar positions. Get on Amazon early, he recommends, as early as you possibly can. Build an email list, run a pre-order, and direct every one of those customers to your Amazon listing before any competitor can get established. Reviews in the first weeks matter enormously to how the platform ranks you, and a well-reviewed listing with traction is harder to bury under a flood of fakes.

Beyond that, he's candid: the options narrow fast. Drive organic content to Amazon so the algorithm prioritizes you. Make clear in your listing that you are the original. And then, largely, outcompete. For most small product companies, he said, that's the only realistic path.

He and his brother, a senior software engineer, have begun working on something more ambitious: an automated tool to scan marketplaces, identify infringing listings, generate takedown notices, and file them at scale. It's early, and Ahlstrom acknowledges the same fundamental limitation applies. If a seller is careful enough to avoid explicit trademark use, there may simply be nothing actionable to file.

"There's just nothing you can do," he said, "until Amazon steps up their ability to keep illegal competition from hitting the marketplace."

For now, Tilted is still growing. Ahlstrom and a small team run fulfillment out of a warehouse in American Fork, where they assemble and package the adapters. The polypropylene top piece is manufactured in Utah. The silicone tubing is sourced from China (at one-fortieth the domestic cost). Packaging is split between U.S. and overseas printers. It's a bootstrapped, unglamorous, genuinely useful operation.

And Ahlstrom, who started the whole thing with a $20 cleaning kit and a door-to-door pitch, is not giving up.

"I love competition," he said. "I expected competition. I just thought it would be in the form of competitors with their own product, advertising it their own way, and not companies illegally using our content, pretending to be us, and scamming people outright."

Tilted Spray Co. is based in American Fork, Utah. The company's adapters are available at tiltedsprayco.com.

The conventional wisdom on AI in politics goes something like this: it's coming, it's vaguely terrifying, and whoever controls it wins. What's less discussed is the basic mechanics — what does AI actually do inside a campaign, and who gets to use it?

Becki Wright has a very specific answer to that question. And today, her company Proximity is putting it on the market.

This week Proximity launched its Smart Messaging Engine, the company's first major AI product, and what Wright describes as the first artificial intelligence built specifically for civic and political leaders. It's not a chatbot. It's not a general-purpose assistant with a civics prompt bolted on. It's a purpose-built intelligence layer that monitors news sources, social media, and community conversations in a leader's district, surfaces what constituents actually care about in real time, and then helps that leader respond — fast, in their own voice, on the right channels.

The pitch is direct: "Most campaigns don't lose because they're wrong about the issues," Wright shared in a recent conversation with TechBuzz. "They lose because they connect with communities too late."

The Gap Nobody Fixed

Here's the problem Proximity is trying to solve. A city council member in suburban Utah, a state senate candidate in Georgia, a nonprofit running an advocacy campaign — they all have the same fundamental challenge. Public opinion is everywhere and fragmented. It's in Facebook groups, TikTok comments, local newspaper letters to the editor, and town hall Q&As that nobody transcribed. Traditional polling takes weeks and costs money most down-ballot campaigns don't have. By the time a candidate figures out that their district suddenly cares intensely about a specific infrastructure issue, the moment to respond authentically has passed.

What the best-funded, most sophisticated national campaigns have always had is a continuous read on that sentiment: opposition research teams, comms directors who live on social media, pollsters running weekly trackers. Smart Messaging Engine is, in essence, an attempt to bottle that infrastructure and sell it for $499 a month.

The product lets leaders build a profile, their district, their issues, their competitors, their preferred sources, and then monitors the information landscape continuously. When something moves, the system alerts them, recommends a response strategy, drafts content (emails, texts, social posts, talking points), and connects to Proximity's existing CRM platform for outreach. A human reviews everything before it goes out. That last part is deliberate and emphasized.

The Builder

Wright came to this from inside the industry she's now trying to disrupt. She spent years as a campaign manager, finance director, and political consultant, putting her close enough to the machinery to be frustrated by it. The tools were fragmented, outdated, expensive, and often inaccessible to the kinds of candidates who most needed an edge. Legacy platforms built for big campaigns, sold to small ones.

She launched Proximity's first beta in 2023, starting with what she calls the "system of record," a unified CRM with email, texting, phone banking, events, and fundraising in one platform. The intelligence layer, the AI, was always the plan. It just had to wait for the infrastructure to exist first.

"We were intentional about building those gateways first," Wright said, "and now being able to add that intelligence so that people could utilize it in the most effective way."

That sequencing matters. A lot of AI products in politics have arrived before the plumbing was ready: flashy tools without the distribution, the trust, or the workflow integration to actually change how campaigns operate. Proximity's approach was slower and more deliberate: build the everyday platform first, earn the user relationships, then layer in the intelligence once people were already living inside the product.

The result is a company now operating in more than 30 U.S. states and Canada, with roughly 25% of its customer base in Utah. Wright reports a two-to-one win ratio for candidates using the platform in the last election cycle, a number that, if it holds, is the kind of data point that writes its own marketing campaign.

The Team, and the Technology Challenge

Here's what makes building political AI uniquely hard: the technology is moving faster than any roadmap can contain.

"AI is changing daily," Wright stated. She's not being hyperbolic. The challenge for her engineering team wasn't just building a product, it was building a product that wouldn't be obsolete by the time it shipped.

Their solution was architectural flexibility. Rather than committing to a single model or infrastructure, Proximity designed its backend to swap models as better ones emerge. They are using smaller, focused models for understanding and classifying signals, and larger models for drafting, with an explicit layer of what Wright calls "civic intelligence" trained on the specific context of political communication.

To anchor the AI buildout, Proximity recently brought on Mou Nandi as CTO. Nandi's background spans computer engineering, healthcare, education, and legal tech — she previously helped build AI-powered document search systems at NetDocuments and founded Monere, a Lehi-based health tech company that uses smartphone photos to screen for anemia, as described by SheTech-TechBuzz media interns earlier this year. It's exactly the kind of pre-hype, applied AI credibility Wright was looking for when she brought Nandi on as CTO.

The Smart Messaging Engine also comes with hard commitments on data: customer data stays with the customer and is never used to train external models. In a political context, where the data in question is constituent relationships and opposition research, that's not a small footnote.

Who's Actually Using This

The customer base is more interesting than it might initially appear. Wright describes it as a "political life cycle." That same person might be a candidate today, an elected official in two years, and running a policy organization five years after that. Proximity is building for the whole arc, not just the campaign moment.

Unexpected use cases have emerged. Educational institutions came asking whether they could use the platform. So did advocacy organizations and think tanks. The platform's core functionality — understand what your community cares about, communicate with them efficiently, track whether it's working — turns out to be useful anywhere that leadership involves constituent relationships.

Two current Utah users Wright highlights: Stephanie Hollist, a Republican running for State Senate to represent District 7 (parts of Davis County and as far east as Morgan), and Karli Black, a Democratic candidate running for the Utah House of Representatives to represent District 58. Different parties, different districts, same platform.

That cross-partisan appeal is part of the brand identity Wright is building. Proximity markets itself explicitly as nonpartisan — a platform that levels the playing field, not one that advantages any particular ideology. In the current political climate, that framing requires some maintenance. But it's also genuinely the product's architecture: the system's intelligence is local and constituent-driven, not nationally ideological.

What's Coming Next

The Smart Messaging Engine is only phase one of what Wright is calling Proximity's "intelligence infrastructure."

Phase two, targeted for Q3, goes a step further. Instead of just monitoring what communities are saying and helping leaders respond, it will track how communities respond to the messages leaders actually send. Close the loop. Turn the outreach data back into intelligence. Recommend next best actions: who to reach, when, on what issue, through which channel.

The framing Wright uses is "engagement pipeline." It is less a messaging tool than an operational guide that tells a leader's team exactly what to do tomorrow, based on what happened today.

If that works the way she describes, it starts to look less like campaign software and more like a continuous intelligence operation, the kind of situational awareness that serious political organizations have historically had to build in-house, at significant cost, and that has largely been unavailable to the local races where most political careers actually begin.

The Bigger Question

There's a version of this story that's slightly alarming. AI making political communication more sophisticated, more targeted, more efficient. Wright has clearly thought about this tension, and she addresses it with a phrase worth quoting directly: people see AI as either a savior or a pariah, with nothing in between.

Her argument is that purpose-built, vertical AI is different from general AI applied to sensitive contexts. The human-in-the-loop requirement on every outbound message isn't just a feature; it's a values statement. The nonpartisan positioning isn't just a marketing choice; it's a structural constraint on how the product is built and sold.

Whether that holds as the product scales is a genuine question. But the more immediate reality is simpler: AI is already in politics. It's being used, right now, by campaigns and political operatives and influence operations, with and without guardrails, in the open and in the dark. The question Proximity is actually answering isn't whether AI should be in politics. It's whether the civic leaders who are supposed to represent normal people will have access to the same tools as everyone else trying to influence them.

That's the case Wright is making. And from a small office at Kiln Lehi, Utah, she's building the infrastructure to prove it.

Proximity's Smart Messaging Engine is available now, starting at $499/month per user. A second product in the civic intelligence suite is expected in Q3 2026. Learn more at proximityimpact.com.

See the full video of the interview with Becki Wright here:

GovTech platform serving 2,300+ municipalities eyes next-generation AI transformation with new private equity partner

Govineer Solutions, a fast-growing software and payments platform purpose-built for local governments, has secured a significant growth investment from global private equity heavyweight TA Associates, a deal that signals mounting investor appetite for AI-driven public sector technology.

The transaction, whose amount was undisclosed, will see existing backer Peterson Partners fully exit its stake in the company, while Govineer's management team and employees will retain meaningful ownership going forward. Financial terms were not disclosed, and the deal is expected to close in Q2 2026, subject to customary regulatory conditions.

From Record-Keeping to Decision-Making

At the heart of the announcement is an ambitious pivot. Govineer wants to move well beyond its roots as a software provider for municipal back-office functions and position itself as what it describes as an "AI-powered operating system" for local government, one that doesn't just log activity, but actively guides decisions, automates complex workflows, and delivers predictive insights to administrators.

"We are reimagining how local governments operate by embedding AI at the core of our platform," said Mike Fabrizio, Co-CEO of Govineer. "Our goal is to move beyond systems that simply record activity to systems that actively guide decisions, automate complex workflows, and help municipalities operate with greater speed, accuracy, and foresight."

It is a lofty ambition, but Govineer has some structural advantages that make the pitch credible. The company's platform currently supports fund accounting, utility billing, online payments, personnel management, and community development workflows across more than 2,300 towns, municipalities, and special districts in 43 states. That footprint translates into large volumes of structured, domain-specific government data, precisely the kind of high-quality training ground that effective AI models require.

A Platform Built Through Acquisition

Govineer has been quietly but methodically assembling its GovTech stack through a combination of organic growth and targeted M&A. Key acquisitions include Caselle, a long-established provider of government financial software, and Black Mountain Software, known for its municipal finance and budgeting tools.

Most recently, in March 2026, Govineer acquired TrueBill Solutions, the utility billing software division formerly known as TruePoint Solutions, from GovPath. The Silver Spring, Maryland-based company brings more than two decades of experience delivering cloud-based utility billing, water resource management, and citizen engagement solutions to water districts, municipalities, and special districts nationwide. The addition meaningfully deepens Govineer's capabilities in one of local government's most data-rich operational areas, water and utility management, further expanding the proprietary dataset that will underpin its AI ambitions.

TrueBill continues to operate under its existing brand, ensuring continuity for its customers while benefiting from Govineer's broader platform and resources.

Together, these moves have deepened Govineer's bench of embedded customer relationships and expanded the structured, domain-specific data assets that sit at the core of its AI strategy.

Co-CEO Nate Quinn said the TA investment would accelerate that strategy further. "With TA's support, we plan to accelerate investment in product development, expand our platform, pursue strategic M&A, and advance our AI capabilities to deliver more predictive, actionable insights for our customers."

Why TA Associates Is Betting on GovTech

For TA Associates, a Boston-headquartered firm with more than $65 billion in capital raised since its founding in 1968, the Govineer deal reflects a broader thesis around mission-critical software businesses sitting on proprietary data moats.

"Govineer sits at the intersection of mission-critical workflows and highly structured public sector data, creating a powerful foundation for AI-driven innovation," said Hythem T. El-Nazer, Co-Managing Partner at TA. "We believe the Company is uniquely positioned to lead the transformation of local government operations through intelligent, data-driven systems."

Chris Hong, Senior Vice President at TA, added that Govineer's deep customer relationships and track record of execution made it a compelling partner for the next phase of growth.

The Bigger Picture

The investment arrives as AI adoption in the public sector accelerates, with local governments under mounting pressure to do more with less amid staffing shortages and tightening budgets. Govineer is positioning itself as the infrastructure layer through which that transformation happens — not by selling standalone AI tools, but by baking intelligence directly into the operational software municipalities already rely on every day.

That strategy is reflected in its M&A playbook. Rather than chasing flashy consumer-facing applications, Govineer has systematically acquired deeply embedded, domain-specific platforms — from financial management to water utility billing — that generate the kind of structured, reliable operational data that AI models can actually learn from. The TrueBill acquisition is a prime example: water and utility billing data, generated consistently across thousands of billing cycles and regulatory cycles, represents exactly the type of structured dataset that can power meaningful predictive and automated capabilities for local governments.

With a well-capitalized new partner, a growing proprietary data advantage, and a clear product vision, Govineer looks set to become one of the more closely watched names in GovTech over the next few years.

Govineer Solutions is headquartered in Provo, Utah. For more information, visit govineersolutions.com.

Fortem Technologies has been selected to provide airspace monitoring and safety support for the 2026 Farnborough International Airshow in Hampshire, England— one of the most significant aviation events on the planet. The Lindon, Utah-based company will deploy four of its TrueView® R40 radars and electro-optical/infrared (EO/IR) cameras to track and record every aircraft movement during the show's live flight demonstrations.

Farnborough International Airshow

For those unfamiliar with Farnborough, it's difficult to overstate its importance to global aviation. Held every two years on the grounds of Farnborough International Exhibition & Conference Centre in Hampshire, about 35 miles southwest of London, the show has been a fixture of the aerospace world since 1948. It is the second-largest airshow of its kind on earth, trailing only the Paris Air Show, and routinely serves as the stage where the industry's biggest deals get done.

At the 2024 edition, the numbers were impressive: over 100,000 visitors, 1,500 exhibitors from more than 60 countries, 90 civil and military aircraft, and more than 420 official government and military delegations packed the show's half-million-square-meter site. Commercial aircraft and engine orders totaling $105.8 billion were announced during the five-day event. This figure represents roughly 13 billion pounds of value to the UK economy alone. The 2024 show was the busiest in recent decades, drawing 33% more visitors than the previous edition.

Since its founding, Farnborough has been the debut stage for some of aviation's most iconic aircraft, from Concorde to the Airbus A380 to the Lockheed Martin F-35 Lightning II.

The 2026 show runs July 20-24, and more than 100 companies have already rebooked exhibition space, including Bell, Viasat, Northrop Grumman, and MBDA.

A High-Stakes Airspace Challenge

With dozens of civil and military aircraft conducting flight demonstrations in close proximity, within tightly defined envelopes and restricted zones, Farnborough presents a uniquely demanding airspace management challenge. Event organizers must ensure that every aircraft stays within its approved boundaries at all times, both for safety and compliance. The consequences of a violation, even an inadvertent one, can affect pilots, spectators, and the reputations of aerospace programs worth billions of dollars.

Under the agreement, Fortem will operate its sensor suite as a fully managed service throughout the two-week event, providing continuous real-time visibility into the airspace above the showground. The system will capture detailed data on aircraft movements during each flight demonstration. When the aircraft land, Fortem's data will allow event organizers to brief pilots on any deviations from their approved flight envelopes, a post-flight review capability that adds an important accountability layer to what is already one of the most tightly managed airspaces in civilian aviation.

"Farnborough brings together some of the most advanced aircraft and flight demonstrations in the world, all operating within a highly constrained airspace," said Jon Gruen, CEO of Fortem Technologies. "Our role is to provide clear, continuous visibility into that airspace so organizers can ensure demonstrations are conducted safely and in accordance with defined parameters."

The R40: Fortem's Most Advanced Radar

The centerpiece of the deployment is Fortem's TrueView® R40, the company's most capable radar system and the newest addition to its TrueView® family. The R40 is engineered for dense, fast-moving environments, exactly the conditions at Farnborough, where military jets, commercial airliners, and experimental aircraft may all be maneuvering in the same airspace within minutes of each other.

The R40 is built on Fortem's low SWaP-C design philosophy: small form factor, low power consumption, and deployable at scale without the infrastructure overhead of traditional radar systems. When networked together — as they will be at Farnborough — multiple R40 units create a persistent, high-resolution picture across the full airspace, filling the gaps that conventional systems often miss at low altitudes and in cluttered environments.

A Growing Track Record in High-Profile Deployments

The Farnborough contract is the latest in a string of significant wins for Fortem, which has been on a run of high-profile deployments globally.

Earlier this year, the company was selected by the U.S. Department of Homeland Security as the sole kinetic counter-drone provider for American venues during the 2026 FIFA World Cup — a deployment covering 11 U.S. host cities and expected to protect venues from more than a million international visitors. It's Fortem's second consecutive World Cup deployment, having also served in Qatar for the 2022 tournament.

In April, Lockheed Martin announced a $25 million strategic investment in Fortem as the initial tranche of its Series B financing round, deepening a collaboration focused on integrating Fortem's systems into Lockheed's Sanctum counter-UAS ecosystem. And last year, Fortem secured follow-on orders for more than a dozen complete counter-UAS systems from U.S. allies in Europe and the Middle East, reflecting accelerating global demand as drone threats have reshaped military and security thinking.

The company's AI-powered SkyDome® Family of Systems — which integrates TrueView™ sensors, command-and-control software, and DroneHunter® autonomous interceptors — has been validated in operational deployments across Ukraine, the Middle East, and East Asia. Fortem remains the only company authorized to deploy a drone-on-drone kinetic interceptor in U.S. airspace.

But Farnborough is a different kind of deployment than those combat-adjacent environments. Here, the mission is airspace awareness and compliance monitoring, not threat interdiction — a demonstration that Fortem's sensor technology is as relevant to civilian aviation safety as it is to defense applications.

A Utah Company on the World Stage

Fortem's selection for Farnborough continues a trajectory that has taken a Lindon-based startup into some of the world's most demanding operational environments. Backed by investors including Lockheed Martin, Toshiba, DCVC, AE Industrial Partners, and Signia Venture Partners, the company opened a new 51,000-square-foot headquarters and manufacturing facility in Lindon last June — more than doubling its production capacity for counter-drone systems and radar hardware.

Being entrusted to watch the skies above Farnborough — where the world's most advanced aircraft will be on display before the global aerospace community — is a meaningful validation for Fortem. When the Red Arrows streak across the Hampshire sky this July and billion-dollar aircraft push the boundaries of their flight envelopes, a system built in Lindon, Utah, will be tracking every move.

Fortem Technologies is headquartered in Lindon, Utah. Learn more at fortemtech.com.

Zach Newman, co-founder and CEO of Enzo Health, is on a mission to rescue the caregivers who care for America's aging millions.

There is a number that should stop you cold: 10,000. That is how many early baby boomers turn 65 every single day in the United States. Not every year. Every day. And the healthcare system tasked with caring for them — home health agencies, skilled nursing facilities, hospice providers — is buckling under the weight of what is coming.

Into that gap steps Enzo Health, a two-year-old Utah startup that is applying artificial intelligence to one of medicine's least glamorous but most consequential corners: post-acute care. Co-founder and CEO Zach Newman sat down with TechBuzz recently to discuss the problem with the calm urgency of someone who has looked at the math and decided that urgency, not panic, is the appropriate response. What he has built, and what he plans to build next, says a great deal about where healthcare innovation is actually needed, and where it has been almost entirely absent.

A System Built for a Different Era

To understand what Enzo Health does, you first have to understand what post-acute care providers are up against. When most people think of healthcare documentation, they picture a doctor typing a few notes into an electronic records system after a visit. The reality for home health clinicians is something else entirely.

"The post-acute reimbursement model is entirely different," Newman explained. "It's based on an assessment that's conducted in the home or in a facility. That assessment is done by a clinician. It could be a nurse, a physical therapist, an occupational therapist. And they are checking the patient's functional, cognitive abilities. And they grade the patient based on this assessment."

What follows is not a handful of paragraphs. The documentation produced from a single patient assessment runs to more than 30 pages of questions. Completing it can take a clinician anywhere from two to three hours, often after they have already spent a full day traveling between patients' homes, delivering hands-on care. Many finish the paperwork at their kitchen table, long after their shift has technically ended.

The consequences are predictable and devastating. Clinicians burn out. They leave. Home health agencies recruit and train workers, only to watch them quit three or four weeks in, overwhelmed by documentation demands that no one fully warned them about. Meanwhile, referrals go unmet. Newman cites research suggesting that roughly 50 percent of referrals sent to home health agencies are rejected outright, not because the agencies don't want the patients, but because they don't have the capacity to take them.

The labor shortage is not a staffing problem in the traditional sense. It is, in significant part, a documentation problem in disguise.

The Technology Gap Nobody Talks About

Part of what makes Enzo Health's opportunity so striking is the baseline it is working from. Post-acute care is not just behind the curve on technology; it is a generation behind.

"This segment of healthcare is probably 10 to 15 years behind where the rest of healthcare was in terms of the technology they had access to," Newman stated. "A lot of the software that was built in this industry was done in the 80s, 90s, early 2000s."

In an era when AI tools are reshaping knowledge work across virtually every industry, home health clinicians have largely been left clicking through ancient interfaces, manually entering answers to assessment questions they have answered hundreds of times before, in systems that seem designed to punish efficiency rather than reward it.

Newman's team arrived in this space not with incremental improvements to existing tools, but with a fundamentally different vision: what if the documentation almost wrote itself?

Enzo's flagship Scribe product, launched about a year ago, does something elegant in its simplicity. When a clinician visits a patient, they open the Enzo application on their tablet and record the visit. The conversation between clinician, patient, and any family members present is captured and transcribed. That transcript is then combined with the referral packet sent by the referring physician, and AI is used to determine how the lengthy patient assessment should be answered — filling out the entire document on the clinician's behalf.

The clinician reviews the output, makes any adjustments, and submits it for compliance review. A process that once consumed two to three hours has been compressed to 20 to 25 minutes.

The numbers are significant. The human dimension may be more so.

Looking Up

There is a moment Newman describes that cuts through the statistics and the reimbursement models and the regulatory complexity to something more fundamental about what healthcare is supposed to be.

"For the first time, clinicians were actually able to look up and observe their patients versus being focused on trying to get their documentation done."

That sentence deserves to sit with you for a moment. These are skilled professionals, nurses, physical therapists, occupational therapists, who have trained for years to observe human beings, to notice the subtle signs that something has changed, that something is wrong, that something needs attention. And the documentation burden of their industry had effectively stolen that capacity from them, forcing them to stare at a screen while a patient sat across the room.

Enzo is giving it back. Clinicians using the Scribe product report noticing things they hadn't noticed before. They feel present in a way that the job had not previously allowed. The ripple effects, on care quality, on clinician satisfaction, on patient outcomes, are difficult to quantify but easy to imagine.

This is what genuine innovation in healthcare looks like: not a flashy consumer app, but a tool that restores something human to people who had nearly given up on finding it in their work.

Fraud, Compliance, and the Integrity of the Industry

No conversation about AI in post-acute healthcare documentation can avoid the subject of fraud. The industry has attracted bad actors who have exploited documentation complexity and reimbursement structures to file false claims, drain Medicare and Medicaid funds, and generate headlines that cast a shadow over providers who are doing exactly what they are supposed to do.

Newman addressed the issue directly, and his framing is worth noting.

"The real providers of care hate fraud," he said. "It makes their lives more challenging because there ends up being more regulation, more controls. So the real providers of care want the fraudsters to be caught."

This is a point that gets lost in coverage of healthcare fraud: the victims are not only taxpayers and the federal government. They are the legitimate agencies that must navigate increasingly burdensome oversight as a result of bad actors' conduct, and ultimately the seniors who depend on those agencies for care.

Enzo has built compliance into its core, not as an afterthought. The company employs clinicians internally, including domain experts who have worked in home health for years, specifically to ensure that the AI's outputs meet a high standard of accuracy and regulatory alignment. The system is designed to assist clinicians in doing their work correctly, not to automate corners that shouldn't be cut.

Newman is clear-eyed about the limits of technology here. Tools can be designed with integrity; they cannot force integrity on those determined to act without it. But for the overwhelming majority of providers who simply want to do their jobs well and get paid fairly for doing so, Enzo is removing the friction that makes honest work harder than it needs to be.

What Comes Next

Enzo Health now serves more than 5,000 end users across home health agencies, with a growing footprint in Utah and nationally. The company has around 45 employees and recently secured new funding it plans to deploy in two directions: expanding into hospice and skilled nursing facility verticals, and helping customers consolidate the fragmented technology stacks they currently manage.

"No customer wants to have a tech stack of seven to eight different tools that they're having to work out of just by switching tabs," Newman says. The vision is a more unified platform that addresses the full arc of the post-acute workflow — admissions, documentation, quality assurance — rather than forcing providers to stitch together point solutions.

The name Enzo, Newman admits, has two origin stories. The real one involves a list of popular dog names. The preferred one invokes Enzo Ferrari, and a company obsessed with speed, precision, and excellence. Whether by canine inspiration or Italian engineering legend, the name has stuck — and the mission behind it is serious.

Ten thousand people turn 65 today. And tomorrow. And every day after that. The caregivers who will meet them at the door deserve tools that honor the weight of that work.

Enzo Health is betting that the technology to build those tools already exists — and that the only thing missing was someone willing to actually build them.

The Investors Taking Notice

The broader investment community is beginning to recognize what Newman and his team have built. As covered by TechBuzz, Enzo Health recently closed a $20 million Series A led by N47, a Palo Alto-based venture firm, bringing the company's total funding to $26 million. Existing investors also participated in the round, including Gradient, the Google-affiliated investment firm based in Palo Alto; Tandem Ventures out of Draper, Utah; and Rigby Watts of Millcreek, Utah.

The backing reflects growing conviction that post-acute care's long-ignored technology gap represents one of healthcare's most compelling investment opportunities — and that Enzo Health is positioned to define the category.

Enzo Health is headquartered in Lehi, Utah. Learn more at www.enzo.health

There is a moment every public speaker knows. The moment the red light on a camera blinks on. Something shifts: posture straightens, sentences get a little more deliberate, the voice finds a slightly different register. It happens to politicians, professors, and pundits. A new study from researchers at BYU Law and NYU now raises a pointed question: does it happen to federal judges, too?

The answer, according to a new paper published in The Journal of Law & Empirical Analysis, is not a clean yes — but it may be closer to yes than the federal judiciary has long assumed.

The study examines what happened inside the U.S. Court of Appeals for the Ninth Circuit, one of the nation’s most prominent federal appellate courts, when it gradually rolled out video recording for oral arguments. Using AI-assisted speaker identification to analyze hundreds of courtroom exchanges across a hand-built dataset of Ninth Circuit proceedings, the research team found preliminary evidence that cameras can alter the dynamics of oral argument. In some analyses, judges spoke more and appeared more likely to interrupt attorneys when proceedings were video recorded rather than audio only. Attorneys, too, seemed to adjust.

But the paper is notably careful in how it presents those findings. Some statistical models found no significant camera effect at all. The stronger evidence emerged in narrower specifications, especially when the researchers compared judges who appeared in both audio-only and video-recorded hearings, and when they allowed the effects of cameras to vary by judicial characteristics. The result is not a definitive verdict that judges perform for cameras. It is something both more modest and more important: the first serious empirical evidence that cameras may change how federal judges behave in court.

For a country still debating whether to put cameras in its highest courts, that is a consequential finding.

The team and the technology

The study is a collaboration between Aaron Kaufman of NYU (Abu Dhabi campus), who conceived the project and developed the underlying AI methodology, and BYU Law professors Dane Thorley and Lucy Williams.

Thorley is an empiricist who has studied judicial behavior for years. Williams is a legal theorist whose work explores rhetoric and judicial action. Together, they brought complementary lenses to a question at the intersection of law, political science, and artificial intelligence.

The technical backbone of the study is a process known as diarization: the automated identification of who is speaking, and for how long, in an audio or video recording. Kaufman had been refining machine-learning methods to apply diarization to court proceedings, where the raw materials are often messy, speaker labels are absent, and large-scale hand coding is prohibitively labor-intensive.

“He jumped on the AI bandwagon pretty early,” Thorley said of Kaufman in a recent interview with TechBuzz News. “He’s done a good deal of work on the use of AI in a number of areas, but particularly in the area of diarization.”

In this study, the researchers used a method called Reference-Dependent Speaker Verification to distinguish judges from non-judge speakers and to identify which judge on a panel was speaking at a given moment. That allowed them to measure how much of each hearing was occupied by judges, how frequently judges spoke, how long their speaking turns lasted, and how attorney speech changed in response.

The AI did not determine whether a judge was fair, theatrical, or persuasive. It served primarily as a measurement tool, helping the researchers identify speaker turns at a scale that would have been difficult to achieve manually. Even so, the process remained labor-intensive. The final dataset included 156 Ninth Circuit cases — 86 audio-only and 70 video-recorded — involving 17 judges and more than 17,000 distinct utterances.

That is substantial for a hand-built empirical legal dataset, but it is still only a slice of the court’s full docket. The sample was also limited in part by the practical demands of the diarization method, which required sufficiently clean reference audio to train the model on particular judges’ voices.

The Ninth Circuit experiment

The Ninth Circuit drew the researchers’ attention for a specific reason: it offered something rare in legal empirics, a real-world policy transition that approximated a natural experiment.

The court did not switch to video all at once. Beginning in 2014, it gradually installed camera equipment in courtrooms across its courthouses, recording arguments on video as those rooms came online. By 2015, video recording had become standard throughout the circuit. During that transition period, some arguments were captured only in audio, while others were recorded and streamed on video.

According to court administrators, cases were assigned to courtrooms in the ordinary course of scheduling, rather than selected for cameras based on subject matter or panel composition. That gave the researchers an opportunity to compare arguments held in camera-equipped courtrooms to those held in rooms without cameras.

The paper describes this rollout as quasi-random rather than truly randomized, and it is careful not to oversell the design. The authors report that the treatment and control groups were not perfectly balanced on all observable characteristics, including some differences in panel composition. Those imbalances matter, and they help explain why the paper’s results are more nuanced than a simple headline might suggest. Still, the court’s staggered transition gave the researchers an unusually strong setting in which to ask whether visibility itself changes courtroom behavior.

What they found

The study’s most attention-grabbing finding is also the one it presents most cautiously: in some analyses, judges on camera appeared to speak more and to interrupt attorneys more frequently than judges in audio-only proceedings.

That conclusion rests on more than one measure. The researchers looked at how much total speaking time judges occupied relative to attorneys, how often judges spoke, and how long individual speaking turns lasted. To approximate interruption, they examined the pattern of attorney speech. In appellate oral argument, attorneys are generally interrupted only by judges, so shorter and more fragmented attorney speaking turns can function as a proxy for greater judicial interjection.

The descriptive trendlines were suggestive. Around the point at which cameras began appearing in Ninth Circuit courtrooms, the researchers observed shifts consistent with judges taking up more of the argument and attorneys receiving less uninterrupted time to speak. In some charts, the change appeared as a clear discontinuity; in others, as a reversal in the existing trend.

But when the researchers applied broader statistical models that pooled all judges together and controlled for judicial characteristics, those effects often disappeared. The stronger findings emerged in more targeted specifications, particularly when the analysis was limited to judges who appeared in both audio-only and video hearings, which allowed for more apples-to-apples comparison. In those narrower models, the camera effect reappeared.

That makes the paper more tentative than many study summaries will likely be. It does not show that cameras always make judges speak more or interrupt more. It shows that there is preliminary, model-sensitive evidence that cameras may do so under certain conditions, and that the effect is strong enough to warrant serious attention.

That nuance matters. It is one thing to say that a camera changes what happens in a courtroom. It is another to say that the evidence is emerging, uneven, and still dependent on how the comparison is drawn. The paper says the latter.

Performance, caution, or both?

The authors situate their work within a broader debate that has long divided judges, lawyers, and scholars. One camp has argued that courtroom cameras promote transparency and public understanding. Another has worried that cameras distort proceedings by making participants self-conscious or performative. The new study speaks most directly to that second claim, what the authors call the “performative judging hypothesis.”

Still, the paper does not and cannot establish motive. A judge who speaks more on camera may be grandstanding. But that same judge may also be trying to clarify the legal issues for a broader public audience, or behaving more carefully because the proceeding is now preserved and viewable. More speech is not necessarily worse speech; more interruption is not necessarily incivility.

Williams emphasized that ambiguity in the TechBuzz interview. Increased talking and interrupting, she noted, is not the same thing as bad behavior. Cameras may induce a kind of disciplined self-consciousness as much as showmanship.

“On the flip side,” she said, “there’s also the possibility that if they know they’re being observed, they behave very cautiously and circumspectly.”

That possibility shadows the entire paper. A courtroom camera is not merely a passive window into judicial behavior. It may also be an intervention that subtly changes the behavior being observed.

Uneven effects across judges

The study also explored whether the effects of cameras varied across different kinds of judges. Here, too, the results are intriguing but tentative.

In interactive models, the authors found evidence that white and male judges became more likely to interrupt when cameras were present, while speaking for shorter durations. Non-white judges, by contrast, appeared to speak for longer and interrupt less in video-recorded proceedings. The paper also suggests that camera effects may interact with the racial, ethnic, and gender composition of attorneys and panels.

These findings are exploratory rather than definitive, and the authors treat them that way. With only 156 cases and 17 primary judges in the dataset, it is difficult to disentangle the independent effects of race, gender, ideology, seniority, and panel role with high confidence. Still, the subgroup patterns are important enough to flag, particularly because recent scholarship has focused on how gender and race shape the back-and-forth rhythm of appellate oral arguments even without cameras.

If cameras do change behavior, the paper suggests, they may not change everyone’s behavior in the same way.

Attorneys changed, too

The effect of cameras did not stop at the bench. Attorneys also appeared to adjust their behavior when arguments were video recorded.

Because the diarization system grouped lawyers together acoustically as non-judge speakers, the attorney-side measures are less granular than the judicial ones. Even so, the researchers found changes in the amount and segmentation of lawyer speech consistent with judges taking up more of the conversational space. That does not necessarily mean attorneys became less effective or less prepared. But it does suggest that cameras may alter the full interactional ecology of oral argument, not merely judicial self-presentation.

That point may prove especially important for future research. If video changes both how judges ask questions and how lawyers answer them, then the courtroom camera is not just documenting a legal exchange. It may be reshaping the exchange itself.

What the study cannot tell us

The researchers are candid about what their paper does not establish.

Most importantly, the study does not tell us whether cameras affect case outcomes. It tracks changes in the conduct of oral arguments, not changes in votes, opinions, or legal doctrine. A judge who speaks more on camera may still decide cases exactly the same way off camera. For some readers, that will limit the force of the findings.

“You could argue, who cares if the judges are taking up more of the airtime during oral arguments if it isn’t making a difference on the back end of the case,” Thorley said. “And I think that argument resonates with me a little bit.”

The paper also does not prove that cameras cause “grandstanding” in the colloquial sense. It identifies changes in speaking patterns, not inner motivations. Nor does it show that the Ninth Circuit’s experience will necessarily generalize to the Supreme Court, to lower-profile federal circuits, to district courts, or to fully remote hearings. The Ninth Circuit is unusually large, visible, and geographically sprawling, covering a huge swath of the US, including the entire west coast, Hawaii, Guam, and Alaska, as well as Montana and Nevada. Its institutional culture may not be representative.

And then there is the question of scale. The authors worked with 156 cases because high-quality diarization at this level of detail still requires substantial front-end labor. The sample is meaningful, but not comprehensive. The paper is best understood as an opening empirical intervention rather than a final word.

A field long ruled by intuition

Even with those limits, the paper marks a real advance in a debate that has often run on assumption more than evidence.

For decades, much of the empirical work on courtroom cameras has relied on surveys: judges, lawyers, and court staff reporting whether they believe cameras changed anything. Those studies generally found that participants perceived little effect. This paper takes a different approach. Rather than ask judges how cameras affect them, it examines what they actually did in arguments recorded under different conditions.

Williams described the project in those terms: less a final answer than a first serious attempt to observe, rather than simply ask, whether cameras alter real judicial behavior.

Thorley recalled that at least one judge he encountered off the record had no trouble believing the premise. “This judge laughed and said, ‘Oh yeah, I totally ham it up when there’s a camera in the room,’” Thorley reported. “This judge was insistent that it didn’t change anything important ... but yeah, I’m probably more performative.”

That anecdote is not the study’s evidence. But it does capture the tension at the center of the paper. Behavioral changes can feel minor to the person making them and still matter in the aggregate, especially when the people involved are federal judges.

The broader stakes

The question of cameras in federal courtrooms has been debated for decades, and it remains unresolved. Most federal district courts still prohibit them. Most federal appellate courts still do not provide routine video of ordinary oral arguments. The Supreme Court remains especially resistant.

That caution sits uneasily with the modern expectation of public visibility. The pandemic forced courts, like nearly every other institution, to become more comfortable with remote participation and digital access. Public demand for greater courtroom transparency has only intensified in the years since.

The issue is not theoretical. In Utah, a judge ruled just days ago that cameras will remain in the courtroom during hearings for Tyler J. Robinson, who is charged in the fatal shooting of conservative activist Charlie Kirk that occurred on UVU's campus last year. Robinson’s lawyers argued that televised proceedings risk turning the case into “entertainment or sensationalism” and could prejudice future jurors. Media organizations and prosecutors pushed back, arguing that open coverage serves transparency and that responsible courtroom reporting is preferable to rumor and commentary outside the courtroom. The dispute arises in a starkly different setting from the Ninth Circuit study — a state criminal trial rather than a federal appellate argument — but it reflects the same underlying tension: whether cameras primarily enhance public access, or whether they also change the proceeding they are meant to reveal.

If anything, the stakes are often higher in trial courts, where cameras may affect not just judges and lawyers but witnesses, victims’ families, and the jury pool itself.

Proponents of cameras argue that they widen access, improve public understanding, and bring the judiciary closer to the democratic values of openness that surround other branches of government. Critics worry that transparency can come at the cost of distortion, that once proceedings are recorded, streamed, clipped, and circulated, participants may begin responding not just to one another but to an imagined audience beyond the room.

This study does not resolve that tension. But it sharpens it. It suggests that camera policy should be understood not merely as a question of public access, but as a question of institutional design. Cameras may make courts more visible while also subtly changing how judges and lawyers behave within them.

That is not necessarily an argument against cameras. It may, however, be an argument for thinking more carefully about how they are introduced: whether proceedings are livestreamed or posted later, whether camera setups are unobtrusive and standardized, and whether rollouts are structured in ways that can be studied rather than simply assumed harmless.

What comes next

The researchers already see several avenues for expanding the work. Newer diarization tools are faster and more scalable than the methods available when this project began, which could allow much larger sets of Ninth Circuit arguments to be processed in the future. Thorley has suggested that expanding the sample could substantially improve statistical precision and help determine whether the apparent camera effects hold up or fade as the data grows.

The authors are also watching other courts. If another federal circuit moves toward video recording, they hope to be involved before the rollout begins, when a cleaner study design might be possible.

“We would love to be involved in figuring out a way to roll it out so that we can study it systematically,” Thorley said.

That ambition points to something larger than a single paper. Courts considering expanded video access now have, at minimum, an empirical warning label. The question is no longer whether cameras in court might affect behavior in theory. There is now evidence — preliminary, mixed, carefully qualified, but real — that they may.

For decades, the cameras debate in federal courts has been sustained by intuition, anecdote, and constitutional principle. This study does not end that debate. But it does begin to move it onto firmer empirical ground. And in a legal culture that asks everyone else to follow the evidence, that is no small thing.

To read the full study by Aaron Kaufman, Dane Thorley, and Lucy Williams, published on February 14, 2026, click here: journals.sagepub.com.

AI experts from UVU turned a room full of veteran auditors into a hands-on AI workshop, and sent them home with a blueprint.

Fifty-two internal auditors from across the Utah System of Higher Education gathered at Utah Valley University on May 5 expecting a briefing. What they received was a masterclass in practical artificial intelligence, delivered by two university interns young enough to have grown up alongside the technology they were teaching.

Mohamed Maiga and Ivan Diaz spent roughly an hour demonstrating not just what AI can do in audit environments, but specifically how to build, deploy, refine, and validate AI agents using tools most professionals already have access to. By the time they finished, the auditors in the room, some with thirty years of experience, had a working blueprint for rethinking how they do their jobs.

The Human Argument Comes First

Recent UVU graduate, Mohamed Maiga, opened the session by making a case that had nothing to do with software.

Drawing on a phrase from outgoing UVU President Astrid Tuminez, "We are human beings, not human doings," he argued that audit-heavy professional environments trap their best people in repetitive manual tasks that crowd out the judgment, collaboration, and relationship-building that organizations actually need most.

"To allow collaboration, you must have time for the human," Maiga told the room. "That's why I love AI. It gives us that power."

The argument carried particular weight coming from Maiga. Originally from Mali, he moved to Utah at seventeen in pursuit of education and opportunity, earned a cybersecurity degree from UVU earlier this month, served as president of the International Students Council, and interned for the vice president of student affairs. He did this all while developing a focused expertise in AI-driven automation. He speaks English, French, and Bambara. His conviction that technology should empower people rather than replace them is not a corporate talking point. It is a personal philosophy, built across continents.

Before presenting a single tool, he made one request of the room: throughout the presentation, identify the most manual, most repetitive process in your own workflow. Because that, he said, is exactly where AI belongs.

A Procurement Card Problem the Whole Room Recognized

Maiga grounded the presentation quickly in a case study that landed with the audience. When he asked how many auditors had encountered procurement card misuse in their work, nearly every hand went up. When he asked whether they believed most of that misuse was intentional, most shook their heads.

That gap, between the complexity of institutional policy and the good intentions of people trying to follow it, is precisely where the Kahlert AI Institute saw an opportunity.

UVU's procurement card guidelines span 99 pages. For every purchase, employees must determine whether the expense is permitted under policy. The rules are detailed, spread across multiple documents, and often interpreted inconsistently.

The stakes are high: these are public funds. Questions like Can I buy cake for a VP's retirement? Are work anniversary gifts covered? may seem minor, but they generate constant friction. That friction compounds into the kind of unintentional misuse that lands on auditors' desks.

The solution built at the AI Institute: a custom chatbot trained directly on UVU's procurement documents. Employees can now ask plain-language questions and receive clear, policy-grounded answers in seconds: no document searches, no supervisor calls, no guesswork. The project was developed by Tyler Small, Senior Director of the Kahlert Applied AI Institute, who fed the agent the relevant policy files, wrote detailed instructions defining what the agent should and should not do, and selected a model capable of reasoning through nuanced policy language.

"There's a lot of negligence, but not fraud," Maiga said. "People don't want to read the rules because they have other things to do. This solves that."

Four Steps to Building Your Own Agent

From the case study, Maiga walked the auditors through a framework for building their own AI agents, deliberately simple, deliberately accessible.

He started by clearing away the fear. An "agent," he explained, is simply any intelligence that performs work for you, whether automatically or on request. Every time someone uses ChatGPT, Claude, or Gemini, they are already working with an agent. A custom one is just a focused version, trained on specific documents, given specific instructions, built for a specific purpose.

The framework breaks it into four steps:

Step 1: Name and describe the agent. Define its purpose in plain language. What does it do? For whom? What problem does it solve? Clarity here sets the foundation for everything that follows.

Step 2: Write the instructions. This, Maiga emphasized repeatedly, is the most important step in the entire process. Instructions must specify what the agent should do and what it explicitly should not do. Use plain English. Be specific. The less ambiguous the language, the more reliable the output. For anyone uncertain where to begin, he offered a practical suggestion: use AI to help draft the instructions themselves. "You want to be directing the AI," he said. "You don't want to get lost in the process."

Step 3: Upload documents directly. A common mistake, Maiga warned, is feeding AI a link and asking it to search for information online. This is less efficient for the LLM and increases the risk of inaccurate outputs. Text documents uploaded directly give the model exactly what it needs. Instead of focusing as much on the search each time, it can focus on interpreting the documents that are already there

Step 4: Select a thinking model. Not all AI models perform equally on complex tasks. For high-stakes applications like internal audit, where an incorrect output could inform a flawed finding, choosing a model with genuine reasoning capabilities matters. Maiga pointed to OpenAI's recent releases as benchmarks, and cautioned against locking into any single tool as the field continues to evolve rapidly.

He closed the framework with a reality check: creating the agent is roughly five percent of the work. The remaining ninety-five percent is iteration — testing outputs, adjusting instructions, and refining the agent's performance over time. Like any professional skill, it improves with deliberate practice.

Ivan Diaz: A Live Demonstration of What "Better" Looks Like

Where Mohamed Maiga built the framework, Ivan Diaz showed it working in real time.

Diaz is an accounting student at UVU with his sights set on a career in public accounting. His project addressed a pain point familiar to anyone who works with institutional financial data: taking raw expense information from nine departments, each with several funding indexes, spanning multiple fiscal years, and producing a clean, formatted, readable report. Done manually, it consumed hours. So he automated it.

Diaz built his agent in Microsoft Copilot, starting with the basic version available free to UVU students and employees. His development process became its own case study in iterative AI improvement, and an honest account of how the best results rarely come from the first attempt.

He began using Copilot's built-in tools to generate initial prompts but quickly hit the ceiling of the platform's default model. Rather than stopping there, Diaz used Gemini Pro to refine his prompts, iterating over one to two weeks and feeding improved versions back into Copilot. When he discovered that Copilot allowed users to switch the underlying model powering the agent, he made the move to OpenAI's GPT 5.4. The improvement was immediate and significant. Output reliability increased sharply, and Diaz discontinued his use of external refinement tools entirely, working instead directly with the agent to diagnose errors and sharpen instructions on the fly.

"The right model changes everything," he told the auditors. He recommended GPT 5.4 specifically for structured, complex data tasks, and encouraged the room to experiment rather than default to whatever model loads automatically.

One practical note for auditors considering similar projects: Copilot's basic version accepts up to three file inputs per session. The premium version supports up to twenty — a meaningful difference for anyone working across multiple departments, fiscal years, or funding indexes simultaneously. Diaz plans to demonstrate the premium version's full capacity at an upcoming UVU conference.

The Step That Makes AI Trustworthy: Validation

The most technically rigorous element of Diaz's presentation was also, perhaps, its most important for an audience of auditors: what happens after the agent runs.

Diaz developed what he calls a run meta sheet. It is a structured validation log that records, for every agent operation, the timestamp, the files used, the file name, the fiscal year, the index, and the number of rows parsed from the source data. After each run, the row count in the meta sheet is cross-referenced against the original CSV files to confirm the agent processed exactly the data it was supposed to process.

In his live demonstration, Diaz showed the agent had parsed 524 rows from the source file. He pulled up the original CSV. The count matched precisely.

"Validating AI outputs is crucial for forming reliable opinions," Diaz stated. "If you can't verify what the agent did, you can't trust what it produced."

For a room of internal auditors, professionals whose entire discipline is built on documented, verifiable evidence, the validation framework was perhaps the session's most resonant idea. AI is not a black box that produces answers to be accepted at face value. It is a tool whose outputs can and must be tested, documented, and confirmed.

The run meta sheet approach carries an additional benefit: transparency. The agent displays the Python code it generates during data processing. Users do not need to know Python to work with the agent effectively, but for those who want to understand precisely what the system did, or need to document the methodology for audit purposes, the visibility is there.

Diaz also addressed data security directly. During the period when he used Gemini and Claude for prompt refinement, no confidential or sensitive information was shared with those external services. The prompts contained only instructions — no files, no financial records, no personally identifiable information. For institutions with strict data governance requirements, the development approach was designed with compliance in mind from the start.

AI Literacy Is No Longer Optional

Neither Maiga nor Diaz stated it quite so plainly, but the subtext running through the entire session was clear: for internal auditors, AI literacy is becoming a core professional competency, not a supplementary skill, not a pilot program, but a fundamental capability with direct implications for the quality and efficiency of audit work.

"There's a real need in this profession for AI literacy," Diaz said. "Being able to understand it, run it, and use it to work more efficiently."

Maiga framed it from the other direction, returning to the idea that had opened the session: AI is most powerful not when it replaces professional judgment, but when it removes the manual burden that prevents professionals from exercising that judgment in the first place.

"The relationships you build are what matter," he said. "AI gives you the time to build them."

For fifty-two auditors who drove from across the state of Utah to be in that room, the message was the same: the manual work piling up on your desk has a solution. And two UVU students just showed you how to build it.

Mohamed Maiga (recent UVU graduate) and Ivan Diaz (Accounting Major) presented an AI Agent Workshop for 52 USHE Auditors, as part of an audit conference at UVU on May 5, 2026. UVU AI students will provide a follow-up demonstration using Copilot's premium version — capable of processing up to twenty files simultaneously — at upcoming UVU conference.

Learn more about the Kahlert Applied AI Institute here.

The 2026 American Causal Inference Conference (ACIC) runs May 11–14 at the downtown Marriott, drawing researchers from MIT, Stanford, Harvard, Johns Hopkins, and institutions across Europe, Asia, and other parts of the world — with a Thursday afternoon networking event open to Utah's broader tech community.

The 2026 American Causal Inference Conference arrives in Salt Lake City next week, bringing together hundreds of researchers, students, and practitioners for four days of intensive work at the intersection of mathematics, statistics, computer science, economics, and medicine.

The conference runs Monday, May 11 through Thursday, May 14 at the Salt Lake Marriott Downtown at City Creek, organized by the Society for Causal Inference, the international organization founded in 2020 whose mission is to foster the science of causal inference and connect the disparate fields that rely on causal knowledge, bridging academic research with policy and practice.

For Utah's tech and data science community, the timing carries significance beyond the geography. Brian Knaeble, a mathematician and professor of computer science at Utah Valley University, has just been elected incoming president of the Society, a role that will connect Utah with a rapidly expanding global field over the next three years.

For those who want to engage without four days of doctoral-level mathematics, Knaeble has reserved space for a networking event on Thursday, May 14, from 1:00 to 3:00 p.m. in the Solitude Room at the Marriott — immediately after the conference closes.

From Prediction to Causation

Most AI systems operating today are built around one core capability: prediction. They identify patterns in historical data and forecast what is likely to happen next. Powerful — but fundamentally different from understanding why something happens, or what would change if you intervened.

Causal inference is the mathematical and scientific pursuit of those harder questions. Researchers in the field describe the problem through what computer scientists call the Ladder of Causation — a three-level hierarchy:

The argument gaining momentum in academic and financial circles alike is that AI systems must climb this ladder to become genuinely capable of guiding real-world decisions, not just forecasting outcomes.

Three major research communities have been arriving at this conclusion from different directions. Researchers in medicine and statistics, particularly at Penn Medicine and Wharton, developed causal methods to evaluate clinical trials — determining not just whether patients taking a drug tend to survive longer, but whether the drug causes improved outcomes. The Society for Causal Inference grew out of those and other efforts and was founded in 2020.

Economists have long relied on causal methods to evaluate policy, and the field received significant validation recently when the Nobel Prize in Economics was awarded to three researchers for foundational causal inference work.

In computer science, Judea Pearl at UCLA received the Turing Award — the field's highest honor — for formalizing the mathematical framework of causal reasoning for AI systems.

At the conference there will be experts in causal representation learning, a technique that can be used to edit images. That technique has performed well in practice, and it is being implemented in modern AI.

Key areas of application for Causal AI include supply chain, manufacturing, healthcare, life sciences, finance, risk management, marketing and customer experience.

The annual ACIC conference is where these threads converge.

The Conference: Scope and What to Expect

ACIC is one of the largest causal inference conferences in the world. Parallel regional events include the European Causal Inference Meeting, held at Oxford this year, and the Pacific Causal Inference Conference, held in China. The Salt Lake conference draws international attendance — researchers from Europe, Asia, and other parts of the world.

The programming is rigorous. Most sessions are pitched at a doctoral level in mathematics or statistics. Monday, May 11 features introductory short courses for those seeking foundational exposure before the main sessions begin. The main conference runs Tuesday through Thursday, May 12–14.

Utah will have a strong presence. More than 40 UVU faculty, graduate students, and industry alumni, spanning mathematics, computer science, statistics, and information systems, have confirmed attendance. Several are already applying causal inference methods in industry and are coming specifically to connect with the global research community working on the same problems.

A Utah Voice in the Field

Dr. Knaeble's path to the society's presidency began when he learned the conference was coming to Salt Lake City. His background is in pure mathematics — he studied at the University of Utah — and his research had led him independently toward causal inference. When he saw the conference location, he reached out to the program committee, shared his work, and offered to help locally.

The society placed him on the program committee. He began recruiting Utah-based attendees.

When the current society president nominated him to run in the global election, Knaeble won — helped, he suspects, by the momentum of a local community invested in seeing the conference succeed in Utah.

He now enters a three-year arc: training in as president-elect starting this May, serving as president from May 2027 to May 2028, and guiding the transition to the next president through May 2029.

"Utah is a wonderful community with a thriving economy and many tech workers with interests that converge on causal inference, and there is an interdisciplinary group of world-class academics about to descend on Salt Lake City. There may be opportunities for collaboration," Knaeble said.

Thursday, May 14: Networking Event to Utah's Tech Community

Knaeble is clear that the main conference is not designed for a general business audience — the talks are dense with mathematics, and attendees without a strong quantitative background are unlikely to get much from the sessions themselves. His recommendation for that audience: skip the conference and come Thursday afternoon.

On Thursday, May 14, from 1:00 to 3:00 p.m., the Solitude Room at the Marriott is open for a networking session — no registration, no cost, no prerequisite. The gathering is specifically designed for Utah professionals in data science, AI, analytics, manufacturing, healthcare, finance, and autonomous systems who want to understand where causal AI is heading and meet researchers who just spent four days on exactly that question.

"The more mathematically focused you are, the more you'll get out of the full conference," Knaeble noted. "But if you're a technical leader or founder desiring some insight into how causal inference may shape the development of future technologies — come Thursday."

Why This Moment Matters for Utah

Causal AI is moving from academic discipline to market reality. Industry analysts project the sector growing at roughly 50 percent annually over the next decade, with major technology companies investing and startups focused exclusively on causal AI platforms attracting significant funding. Applications are emerging in manufacturing — where causal reasoning enables genuine root cause analysis rather than defect prediction — as well as healthcare, logistics, finance, autonomous systems, and defense.

For Utah companies already working in AI and data science, the conference represents a rare proximity to the global research community driving these advances.

Companies interested in sponsorship or exhibitor opportunities at ACIC should contact Knaeble directly. Through his position on the program committee, he can connect interested parties with the right people.

American Causal Inference Conference

May 11–14, 2025 | Salt Lake Marriott Downtown at City Creek

Conference Program

Thursday Networking Event — Free and Open to All

May 14, 1:00–3:00 p.m. | Solitude Room, Salt Lake Marriott Downtown at City Creek

To connect with Brian Knaeble regarding the Thursday event, sponsorship, or the society: bknaeble@uvu.edu

To learn more about the Causal Inference field, read this introductory article.

Additional reading: Answering Causal Questions Using Observational Data:

"Just try it, and don't be intimidated if you don't know something," encourages Abbey Lasater, a rising leader in data engineering and mathematics.

Lasater is the Associate Vice President of Data Engineering at Sunwest Bank in Sandy, UT, a role she stepped into in October 2025. Sunwest is a bank that specializes in working with small-medium businesses, privately held corporations, family offices, and real estate developers and investors. Attracted to the community focus and the opportunity to explore data engineering, she is the first data engineer at the bank, with major projects already underway. Her current focus is unifying the bank's data onto a cloud platform, where she can continue to explore and understand data structure and usage in a banking context.

Lasater's initial exposure to data came during her first internship at Swire Coca-Cola, where she was new to analytics but enthusiastic to try something new. As an intern at Swire Coca-Cola, she was tasked with building an AI chatbot, analyzing inventory, and working on supply chain and logistics projects.

After the internship, she was brought on full-time as a data scientist at Swire Coca-Cola, where she was soon given the opportunity to mentor the next intern to follow in her footsteps.

"Coming in, they didn't have an established data team At Sunwest Bank. I'm the first data engineer at Sunwest Bank, so the data wasn't what I was used to," she explains. "Nobody was there to just give me a knowledge transfer on what they had done in data engineering, it was my job to come in, explore the data, and understand what needs to be done. Especially being in tech, things change so much, so quickly. There's so much innovation. You're never going to know everything."

She lives by that advice, too. Lasater's path through school wasn't always straightforward — she initially struggled with computer science and even considered dropping it entirely. She switched her major to mathematics, a subject she felt more comfortable with from high school, reasoning that computer science wasn't for her since she'd never taken programming classes before. But after settling into math, she decided to add a computer science minor. "It was kind of a full circle moment," she describes. "Like, 'I can do this'. This isn't as hard as I thought it was going to be, and I ended up loving it." She is currently pursuing her master's degree in computer science from Utah Valley University, reflecting her commitment to growth within her career.

Lasater is passionate about inclusivity in tech, an interest that grew from the lack of diversity she observed in her own classes. "I had never seen a lot of women in the tech space, or a lot of my math classes were male dominant… I just hope that these young women who are growing up now can see that it's just as much for them as it is for anyone else." She also wants young women to know: "Just try it and don't be intimidated if you don't know something. The fun part about tech is the fact that you don't know everything."

Through the challenges of balancing her personal, professional, and academic life, Lasater has learned the importance of prioritization and having realistic expectations. She also wants young women to find their own approach to learning and to feel comfortable not having all the answers.

From a young woman drawn to math and technology to the first in her family to attend college and earn a bachelor's degree, Abbey Lasater's journey is one of perseverance and growth. Advancing from intern to employee to mentor, she has continued to develop her skills and leadership, now bringing that same energy to Sunwest Bank.

Her journey shows young innovators that success doesn't come from having all the answers, but from the willingness to adapt and learn. For anyone considering a future in data science, tech, or simply a new opportunity they're not sure about, her advice is simple but powerful: "Just try it, and don't be intimidated if you don't know something."

Siya Jain is a student at Cedar Valley High School in Eagle Mountain, Utah, and has been active in SheTech for over two years. Aspiring to become an entrepreneur, she develops her skills through coursework, leadership roles, and hands-on experiences, including networking with local founders and operators. Siya serves as SheTech president at her school and is a member of the program’s student board, helping to lead and represent her chapter.

Also from Cedar Valley High School, Izzie Larson is a senior pursuing an Associate’s Degree from UVU. She has a strong interest in linguistics, hydroponics management, and business communications, and is passionate about impromptu and public speaking.

Through the SheTech Media Internship with TechBuzz News, Siya and Izzie interview and write about Women Tech Awardees. Their reporting has been published on TechBuzz News, Silicon Slopes, and other media outlets.

A new international partnership could put rural Utah at the center of a cleaner, more strategic critical minerals supply chain.

Milford Mining Company Utah (MMCU) announced it has signed a memorandum of understanding with Osaka City, Japan-based Furnace Japan Co. to pilot advanced processing technology designed to recover tungsten and other critical minerals from historical mine tailings while dramatically reducing environmental impact.

The agreement pairs MMCU’s mining assets in Beaver County, including copper resources, historical tungsten deposits and legacy tailings, with Furnace Japan’s patented electric resistance furnace (ERF) technology. The companies say the collaboration could transform previously discarded mining waste into commercially viable domestic mineral supply.

The announcement marks another major step in MMCU’s evolution after what the company previously described as a transformative year that included restarting copper operations at one of the highest-grade copper mines in the United States. The company has increasingly positioned itself as both a domestic copper producer and a broader critical minerals platform focused on U.S. supply chain security.

Unlike traditional extraction systems that rely heavily on acid-intensive processing, Furnace Japan’s pyrometallurgical ERF technology uses primarily electrical energy to process ore and mine tailings. The process significantly reduces chemical waste and produces environmentally stable slag that can potentially be reused for industrial applications.

Under the agreement, the companies plan to develop a pilot plant in Milford capable of recovering tungsten and additional valuable minerals from historical tailings sites. If testing is successful, the partners intend to scale the technology into a commercial operation and eventually apply it to newly mined ore as well.

The project arrives as demand for critical minerals accelerates across AI infrastructure, electrification, semiconductors and defense manufacturing. Tungsten, in particular, is considered strategically important because of its use in aerospace systems, electronics, industrial tooling and military applications.

Utah Governor Spencer Cox called the partnership an example of how advanced technology and Utah’s natural resources can work together to strengthen the state’s economy and domestic mineral capabilities.

“The agreement with Furnace Japan highlights Milford Mining Company Utah’s global leadership in pioneering innovation that shows what’s possible when Utah’s resources are paired with world-class technology,” Cox said in a statement.

MMCU Chairman Roger Barris said the company sees the initiative as an opportunity to convert mining byproducts into strategic materials while strengthening American supply chain independence.

The location could become a differentiator. MMCU’s operations sit within Utah’s Milford Renewable Energy Corridor, giving the project potential access to geothermal and wind energy sources that could power future processing operations with a lower carbon footprint than conventional mining systems.

The partnership also reflects broader U.S.–Japan cooperation around critical mineral supply chains as both countries seek alternatives to foreign-controlled processing capacity. By combining Japanese metallurgical technology with U.S. mining assets and infrastructure, the companies say the collaboration could establish a new model for allied industrial development focused on sustainability and national security.

The Milford project also reflects a broader resurgence of copper and critical mineral development across Utah, where several previously dormant mining operations have recently restarted amid rising demand tied to electrification, AI infrastructure and domestic supply-chain security.

In Beaver County, Milford Mining recently restarted copper production at a mine that had sat dormant since 2019, part of a growing trend across Utah where older mining assets are being revived with modern technology and new investment. Separately, Mariana Minerals recently reopened the former Lisbon Valley Copper Mine near La Sal as what the company described as the world’s first fully autonomous copper mining operation, deploying AI-driven systems, autonomous haul trucks and robotic drilling technologies, as covered by TechBuzz. Together, the projects signal how Utah is emerging as a national center for next-generation domestic mineral production.

For Utah, this latest Milford Mining announcement reinforces the state’s growing role in domestic critical mineral production at a time when federal policymakers and private industry are racing to secure reliable North American supply chains for energy transition and advanced manufacturing industries.

Learn more at milfordmining.com and furnace.jp/en.