The magentIQ Show Ep. 12 | The Case for a Council of Agents, the Companies Rehiring the Experts They Replaced, and Why Human Judgment Still Wins

The magentIQ Show Ep. 12 | The Case for a Council of Agents, the Companies Rehiring the Experts They Replaced, and Why Human Judgment Still Wins

Ian Barkin and David Brain get into what happens when you trust a single AI model too much, and what to do instead. A client's agent confidently got a healthcare policy answer wrong, which sets up David's most practical idea yet: a council of agents, several different models cross checking each other so no one blind spot slips through. From there, the quiet trend of organizations rehiring the very experts they let go, and why human judgment keeps proving itself the differentiator. Positive, practical, and a little bit cheeky. (1:13) What's in this episode: AI employees, the council of agents, symbiotes, and a second grade confession (2:25) Founders journal: Nashville, live music, and the managed service providers fielding "where did you host that vibe coded thing?" (5:38) Fable 5 is back: a model released, pulled, and now re-approved, and what that says about depending on any single model (7:55) Tiered access to frontier models: who gets the most capable tools first, and what it means for smaller players (12:01) AI sovereignty and the kill switch: why where a model is governed is now part of your strategy (22:48) AI employees and the accountability gap: how agents are quietly disrupting who owns the work (24:10) Confident and wrong: a major report built on dozens of fabricated sources, and the guy in the pub problem (25:20) Ian's second grade dictionary confession, and why made up facts are so convincing (29:52) The story that started it: a single agent hallucinates on a healthcare policy question (30:35) The council of agents: borrowing from crowdsourcing to have multiple models check each other for accuracy (32:40) Why the council has to be multi vendor, and how it also protects you from a single model outage (37:48) The great rehiring: carmakers, banks, and consultancies bringing back the experts they replaced when AI fell short (41:21) The mediocrity trap: why hiring "good enough plus AI" costs you the experts you will need in five to ten years (45:45) From robots to agents to symbiotes: the new label for an old truth, humans and AI working better together Listen now, and tell us where a second opinion would have saved the day. Find us in your favorite streaming service: iTunes: https://podcasts.apple.com/us/podcast/the-magentiq-show/id1896570951 [https://podcasts.apple.com/us/podcast/the-magentiq-show/id1896570951] Spotify: https://open.spotify.com/show/033f2oKxnFr5fmcpdSHjSL [https://open.spotify.com/show/033f2oKxnFr5fmcpdSHjSL] iHeartRadio: https://iheart.com/podcast/333292440 [https://iheart.com/podcast/333292440] Got a perspective worth sharing? We are always looking for guests. Reach out at info@bemagentiq.com [info@bemagentiq.com].

The magentIQ Show Ep. 11 | Why Great Ideas Matter More Than Token Maxing, and the Risk of Betting Your Business on One AI Model

Ian Barkin and David Brain are back, and getting comfortable. This one ranges from World Cup soccer to the real bottleneck in AI, which turns out not to be tokens or building, but knowing what to build. They dig into why great ideas beat raw consumption, why last year's models are quietly the smart buy, and what a sudden model shutdown teaches every operator about not putting all their eggs in one basket. Plus a genuinely new idea about keeping the business running when the model goes dark, involving, of all things, senior centers. (1:06) Founders journal: World Cup soccer, manufactured hydration breaks, and Europeans touring America (8:39) Apple's WWDC, the AI Siri that is still not here, and the chip shortage driving up prices (11:10) Token maxing and return on token: why using more compute just means more to wade through (15:00) Why great ideas matter more than token maxing: the real constraint is imagination, not the ability to build (18:00) The psychology of innovation: how to choose a few ideas and retire the rest without losing your best people (22:15) Why older models are the smart buy: getting it right with last year's Gemini as prices keep dropping (24:51) Commoditization, margins, and where AI pricing is heading (30:00) Fable 5 gets shut off in 72 hours: a frontier model here one moment, gone the next (37:17) The geopolitics of AI: models built on humanity's collective knowledge, and the push toward open source and local (41:12) The ratchet effect and agentic business continuity: what happens when the model your operation runs on goes dark (44:32) The senior center solution: the analog escrow of people who still remember the old way (47:52) Staying model agnostic: why you prototype in the latest and greatest but build so you can switch (50:46) The build update: the imagination platform, the holodeck, and why your co-pilot is now a sheep Listen now, and tell us where your own imagination gap is biggest. Find us in your favorite streaming service: iTunes: https://podcasts.apple.com/us/podcast/the-magentiq-show/id1896570951 [https://podcasts.apple.com/us/podcast/th%EE%80%80e-magentiq-show/id1896570951] Spotify: https://open.spotify.com/show/033f2oKxnFr5fmcpdSHjSL [https://open.spotify.com/show/033f2oKxnF%EE%80%80r5fmcpdSHjSL] iHeartRadio: https://iheart.com/podcast/333292440 [https://iheart.com/podcast/333292440] Got a perspective worth sharing? We are always looking for guests. Reach out at info@bemagentiq.com [info@bemagentiq.com].

The magentIQ Show Ep. 10 | The Biggest AI Model Is Rarely the Right One, AI Budgets Are Not Adding Up, and Speed Is Not a Strategy

Ian Barkin and David Brain go behind the scenes on the agentic platform they have been building, and dig into why the biggest model is rarely the right one, why this year's AI savings are quietly funding next year's budget, and why going faster will not fix a process that was never designed for AI in the first place. A practical, pattern matching conversation from two people who have lived through this cycle before. (0:00) Why there is nothing safe and low hanging in AI transformation, and how big, bold, brash expectations slow you down (0:55) Founders journal: black t-shirts, decision fatigue, GLP-1s, and the great tailor shortage (10:54) The Bain study: AI is not delivering the expected impact, and next year's budget is built on savings that never showed up (16:08) The underpants gnomes problem: use AI, question mark, value is not a plan (21:30) Capturing the thinking, not just the steps: David on building an agentic platform rooted in process re-engineering (28:28) The 10x myth: why companies keep trying to 10x a broken process instead of redesigning it (30:49) Why the forward deployed engineer puts too much on one set of shoulders (38:45) Orchestrating the right model: vendor lock-in, and why you don't need a McLaren to drive down the street (40:31) Small and custom language models: lower cost, lower energy, and good enough for most real work (43:49) Chinese and open source models, the nine month gap, and the regulation that may be coming (48:54) Recursive, self improving models and why a human master architect still has to stay in the loop (53:07) The hype versus reality gap, told through the Siri that won't tell you if it's going to rain Listen now, and send us your questions. We work through this stuff too. Find us in your favorite streaming service: iTunes: https://podcasts.apple.com/us/podcast/the-magentiq-show/id1896570951 [https://podcasts.apple.com/us/podcast/the-magentiq-show/id1896570951] Spotify: https://open.spotify.com/show/033f2oKxnFr5fmcpdSHjSL [https://open.spotify.com/show/033f2oKxnFr5fmcpdSHjSL] iHeartRadio: https://iheart.com/podcast/333292440 [https://iheart.com/podcast/333292440] Got a perspective worth sharing? We are always looking for guests. Reach out at info@bemagentiq.com [info@bemagentiq.com].

The magentIQ Show Ep. 9 | The Velocity Gap: Why AI at Home Races Ahead While AI at Work Stalls (Guest: Phil Fersht)

People can plan a week of meals, diagnose an ailment, and draft a proposal with AI on a Sunday, then walk into work on Monday and find everything still stuck in silos, legacy systems, and tribal knowledge. In Episode 9 of The magentIQ Show, Ian Barkin sits down with industry analyst and icon Phil Fersht to name that dissonance, the velocity gap, and to talk honestly about why so few enterprises are closing it, what real AI transformation actually demands, and why authentic human voice matters more than ever in an age of AI slop. Phil Fersht is widely recognized as one of the world's leading analysts on reinventing business operations to exploit AI and global talent. He is the founder and CEO of HFS Research, the analyst who authored the first report on Robotic Process Automation in 2012, and the voice who coined OneOffice, trademarked the Generative Enterprise, and more recently framed the rise of Services-as-Software. He is also the author of Horses for Sources, the most widely read blog in the global services industry, now in its nineteenth year, and a three time Analyst of the Year. Here is what Ian and Phil unpack: * Why the velocity gap exists, the widening distance between how fast people adopt AI personally and how slowly their organizations actually move, with only a small fraction of the global 2000 making real headway * Why LLMs are not AI transformation, and why meaningful change requires CEO and COO sponsorship and a wholesale rewiring of people, data, and process rather than incremental tweaks * Why authentic voice is becoming a differentiator as AI slop floods inboxes and feeds, and how to use these tools for research and structure without losing the soul of what you write * Why the future of services belongs to nimble, process led, outcome driven firms, and why being close to the operation beats dropping a full stack engineer into a problem The throughline is energizing rather than alarming. The organizations that win will be the ones whose leaders treat AI as a fundamental rethink of how the business runs, stay close to their processes, and keep real human judgment and voice at the center. The opportunity to leap ahead is wide open, especially for the small and nimble. Listen now and tell us how wide the velocity gap looks in your own organization. Resources from Phil Fersht: Podcast, From the Horse's Mouth, Intrepid Conversations with Phil Fersht: * Apple: https://podcasts.apple.com/us/podcast/from-the-horses-mouth-intrepid-conversations-with/id1769638427 [https://podcasts.apple.com/us/podcast/from-the-horses-mouth-intrepid-conversations-with/id1769638427] *  Spotify: https://open.spotify.com/show/7uS2O3N9JdQnpWLT22qy5O [https://open.spotify.com/show/7uS2O3N9JdQnpWLT22qy5O] *  YouTube: https://www.youtube.com/@FromTheHorsesMouthPod [https://www.youtube.com/@FromTheHorsesMouthPod] Blog, Horses for Sources: https://www.horsesforsources.com/ [https://www.horsesforsources.com/] Got a perspective worth sharing? We are always looking for guests. Reach out at info@bemagentiq.com [info@bemagentiq.com].

The magentIQ Show Ep. 8 | Lessons Not Learned: Why Unlimited Tokens Backfire, the Process Map You Cannot Skip, and the Myth That Speed Is a Strategy

The company burning billions on AI is somehow getting less done. In Episode 8 of The magentIQ Show, Ian Barkin and David Brain dig into why, working through the mistakes the AI era keeps repeating even though most of them were supposed to be learned a decade ago. Here is what Ian and David unpack: * Why the era of unlimited tokens is turning, and why disciplined AI is proving cheaper and more productive than handing everyone a blank cheque to use as much as possible * Why choosing the tool before defining the problem still trips teams up, and why a demo is no substitute for knowing where you are and where you actually want to go * Why expecting staff to 10x themselves overnight is a tall order, and why the forward deployed engineer cannot be a whole team of specialists squeezed into one person * Why speed for its own sake is not a strategy, and how the old trade off between time, cost, and quality still holds no matter how fast the tools have become The throughline is encouraging rather than cautionary. The teams that win are the ones that map where they are, agree on the outcome they want, choose the right tool for the job, and give their people the design time and support to do it well. Enthusiasm builds momentum, but pairing it with context and a little hard won history is what turns an AI project into a real result. Listen now and tell us which lesson you see playing out most in your own organization. Find us in your favorite streaming service: iTunes: https://podcasts.apple.com/us/podcast/the-magentiq-show/id1896570951 [https://podcasts.apple.com/us/podcast/the-magentiq-show/id1896570951] Spotify: https://open.spotify.com/show/033f2oKxnFr5fmcpdSHjSL [https://open.spotify.com/show/033f2oKxnFr5fmcpdSHjSL] iHeartRadio: https://iheart.com/podcast/333292440 [https://iheart.com/podcast/333292440] Got a perspective worth sharing? We are always looking for guests. Reach out at info@bemagentiq.com [info@bemagentiq.com].