Speed vs. Trust — How To Ship AI Without Sacrificing Either | EP34

Speed vs. Trust — How To Ship AI Without Sacrificing Either | EP34

AI trust in regulated industries isn't a compliance checkbox — it's an operational discipline you build into every iteration. Abhijeet Gulati, head of AI at Mitchell (an Enlyte company), has spent nine years shipping AI into property and casualty insurance claims management, where accuracy is non-negotiable and a wrong estimate has real financial consequences for real people. In this conversation, Abhijeet introduces the velocity-veracity paradox — the tension between time to market and time to trust — and breaks down how his team resolves it. You'll hear how Mitchell decides when AI is production-ready vs. when to pause, why bias must be operationalized as a routine rather than treated as a one-time audit, and how human-in-the-loop design works in practice when AI confidence meets uncertainty. This is the final episode of our three-part series on AI as an operations force multiplier: EP32: AI Didn't Replace These Workers — It Gave Them Their Mission Back — https://youtu.be/ziCYZtWNzps [https://youtu.be/ziCYZtWNzps] EP33: How HCA Turns Clinical Notes Into Intelligence — https://youtu.be/XAGaIPnuUhY [https://youtu.be/XAGaIPnuUhY] Book an Insight Prism workshop because you'll get a structured framework to identify where AI can create operational intelligence from your existing data: https://www.insight.com/en_US/what-we-do/methodology/insight-prism.html [https://www.insight.com/en_US/what-we-do/methodology/insight-prism.html] Subscribe to Insight On for new episodes every week. #AItrust #RegulatedAI #ResponsibleAI #InsightOn Chapters * 00:00 — Welcome and introduction * 01:57 — What Mitchell does in claims management * 03:23 — How AI powers the insurance claims experience * 05:22 — The velocity-veracity paradox explained * 08:00 — Building and maintaining customer trust with AI * 10:18 — Build vs. buy decisions for regulated AI * 11:53 — What generative AI makes possible now * 13:26 — Framework for deciding when AI is ready to ship * 15:29 — Operationalizing bias as a continuous routine * 18:09 — The one question every AI leader should ask * 20:35 — Closing thoughts

How HCA Turns Clinical Notes Into Intelligence | EP33

Generative AI is making it possible to process unstructured clinical data at enterprise scale for the first time. HCA Healthcare's AVP of Data Science, Sara Liao-Troth, PhD, MBA, explains how her team is extracting intelligence from doctor's notes, nursing handoffs and free-text records that represent roughly 50% of HCA's patient data across 44 million annual encounters. But the conversation takes a surprising turn when Sara reveals how she's rethinking team composition. Rather than relying solely on senior data scientists, she's pairing experienced practitioners with junior talent and interns who push generative AI tools to their limits precisely because they don't know the old way of doing things. The result: a spine surgery data extraction problem solved by an intern that now informs enterprise-wide benchmarking and supply chain decisions. You'll learn how HCA approaches responsible AI in a zero-tolerance-for-hallucination environment, why Sara argues you just start experimenting with AI rather than planning your strategy, and how targeted teams can build internal capability to adapt as the technology keeps shifting. This is episode two of our three-part series on AI as an operations force multiplier. Catch episode one: AI Didn't Replace These Workers — It Gave Them Their Mission Back | EP32 [https://youtu.be/ziCYZtWNzps] Book an Insight Prism workshop because you'll get a structured framework to identify where AI can create operational intelligence from your existing data: https://www.insight.com/en_US/what-we-do/methodology/insight-prism.html [https://www.insight.com/en_US/what-we-do/methodology/insight-prism.html] Subscribe to Insight On for new episodes every week. #HealthcareAI #DataScience #GenerativeAI #InsightOn Chapters * 00:00 — Welcome and introduction * 00:47 — What is HCA Healthcare and Sara's role * 03:17 — 50% of patient data trapped in free text * 05:14 — How generative AI changes the patient experience * 07:11 — Rethinking data science team composition * 09:21 — Will AI eliminate data science roles * 12:56 — Giving interns impossible problems to solve * 14:29 — The spine surgery use case explained * 17:57 — How to plan when AI keeps changing * 19:55 — Responsible AI in healthcare * 21:22 — One thing leaders should believe

AI Didn't Replace These Workers — It Gave Them Their Mission Back | EP32

AI agents for non-emergency calls are solving a problem that policy, process changes, and hiring couldn't fix for nearly a decade. At 911 centers across the United States, the majority of incoming calls are non-emergency inquiries — parking tickets, road obstructions, animal control — handled by operators trained for life-or-death situations. Viiz Communications built conversational AI agents on Google's Contact Center as a Service (CCaaS) platform to intercept those calls, provide full responses, and keep humans focused on emergencies. Chad Brothers, VP of emergency services programs at Viiz, explains how the company recognized a massive problem — skilled 911 operators drowning in work that didn't require their expertise — and built an AI solution that the industry had been waiting years for. You'll hear why one of the most cautious industries in America adopted AI faster than anyone expected, how Viiz proved the concept internally by taking QA coverage from 1.5% to 85% of calls in less than two weeks, and what every organization can learn about protecting skilled workers' time and mental capacity for the work that truly requires human expertise. Talk to an Insight specialist about Insight AI solutions because you'll get a clear path from one operational friction point to measurable AI results — the same approach that worked in this risk-averse industry: https://ips.insight.com/en_US/what-we-do/expertise/data-and-ai.html [https://ips.insight.com/en_US/what-we-do/expertise/data-and-ai.html] Subscribe to Insight On for new episodes every week. #AI #AIagents #PublicSafety #ContactCenter #InsightOn Chapters (5–12) * 00:00 — Welcome and introduction * 02:18 — What Viiz Communications does * 03:02 — Why this AI story flips the job displacement narrative * 04:32 — The 911 staffing crisis explained * 07:32 — 60% of 911 calls aren't emergencies * 09:51 — What Viiz built with Google CCaaS * 11:12 — How emergency calls still reach humans * 12:49 — Why a risk-averse industry adopted AI fast * 15:37 — The parallel to every knowledge worker * 17:35 — How to start with AI in your organization * 19:27 — The QA agent that 3Xed output in days * 22:20 — Homework for every listener

Data Is My AI Strategy" — How UC Riverside Is Outrunning Organizations With Bigger Budgets | EP31

Most organizations spend months gathering requirements before a single line of code gets written. UC Riverside's CIO Matthew Gunkel is compressing that entire cycle into a single day — running live design sessions where stakeholders walk in with a problem and walk out with a working application spec built in real time using AI. In this conversation, Gunkel explains why his AI strategy starts and ends with data — not models, not agents — and how UC Riverside's lack of legacy data infrastructure became an unexpected advantage, letting them skip traditional data warehouses entirely and move straight to vector databases and graph knowledge. He also shares how the university is deploying AI agents for student wellness outreach and procurement, using Notebook LM as a classroom RAG tool, and why the most practical AI skill you can learn right now has nothing to do with code. This is the final episode in our citizen development series. Catch up on earlier episodes with Joseph Schultz [https://youtube.com/watch?v=rf8MyG22FnA&feature=youtu.be] and Jason Dittmer [link pending] for the full picture of how individuals and organizations are building their own solutions. Contact an Insight AI specialist because you'll get solutions tailored to your data maturity, infrastructure, and use cases — not a generic platform pitch: https://www.insight.com/en_US/what-we-do/expertise/data-and-ai.html [https://www.insight.com/en_US/what-we-do/expertise/data-and-ai.html] Subscribe and follow Insight On for new episodes every week. #AI #DataStrategy #HigherEducation #CitizenDevelopment #InsightOn Chapters (8) * 00:00 — Welcome and introduction * 00:38 — Citizen development at institutional scale * 03:06 — Change management at the speed of AI * 04:00 — Live design sessions that replace months of planning * 06:05 — Procurement and student wellness AI use cases * 10:14 — Authentic assessment and AI in education * 13:28 — Why data is the AI strategy * 18:25 — The case for learning folders and markdown

Data Governance Led MERGE to an Unexpected Product | EP30

Most organizations are sitting on fragmented data from dozens of sources with no fast way to normalize it, query it, or get answers. MERGE had the same problem — and what they built to fix it internally became a product their clients needed too. In this conversation, Jason Dittmer, SVP of TechOps at MERGE, explains how his team built an automated pipeline on Google Cloud (BigQuery, Looker, Gemini, Google SecOps) to solve their own data fragmentation problem — cutting normalization time from weeks to minutes. Then they heard the same pain from clients and shipped it as a marketplace product. One healthcare client now runs 30 disparate data sources through the same pipeline. If you're thinking about how to productize your own internal AI work, this breakdown of moving from efficiency to revenue is worth reading alongside the episode: https://www.insight.com/en_US/content-and-resources/blog/from-efficiency-to-revenue-productizing-enterprise-ai.html [https://www.insight.com/en_US/content-and-resources/blog/from-efficiency-to-revenue-productizing-enterprise-ai.html] Jason also breaks down MERGE's "Drink Your Own Champagne" philosophy — the idea that you should prove a solution internally before ever bringing it to a client. He shares what it takes to move past the pilot phase, and how MERGE's Humanity Suite puts the human factor at the center of AI-powered marketing. If you're still sorting out what AI agents can actually do in this context, the AI agent cheat sheet is a good companion: https://www.insight.com/en_US/content-and-resources/guide/the-ai-agent-cheat-sheet.html [https://www.insight.com/en_US/content-and-resources/guide/the-ai-agent-cheat-sheet.html] This is the second episode in a series on organizations building AI solutions from the inside out. In the first, Joseph Schultz at JE Dunn Construction explains how field workers with no coding background are building their own AI-powered tools: https://youtu.be/rf8MyG22FnA?si=bs2jyJ2u_1_PO66N [https://youtu.be/rf8MyG22FnA?si=bs2jyJ2u_1_PO66N] Book an Insight AI Readiness and Governance Workshop because you'll get a clear framework for moving your AI projects from pilot to production: https://www.insight.com/en_US/content-and-resources/solution-briefs/ai-adoption-with-ai-readiness-governance-workshop.html [https://www.insight.com/en_US/content-and-resources/solution-briefs/ai-adoption-with-ai-readiness-governance-workshop.html] Subscribe to Insight On for more conversations with the leaders building what's next. Chapters (5–12) * 00:00 — Welcome and introduction * 01:32 — What MERGE does and what's on Jason's desk * 03:17 — Drink Your Own Champagne explained * 05:06 — The internal data problem that started it all * 08:27 — What made this solvable on Google Cloud * 10:50 — Three months from idea to internal product * 12:14 — Surprising insights from contextually aware data * 13:13 — The Humanity Suite and infinite individualism * 16:07 — Could this have happened a year ago * 17:30 — How AI perception changed inside MERGE * 18:30 — What's next for MERGE * 19:22 — Advice for leaders stuck in the pilot phase #AIDataPipeline #GoogleCloud #EnterpriseAI #AIPilotToProduction #InsightOn