Advertise on podcast: In The Loop

Jev is a new kind of AI model that can't write anything. Instead it does the most important but most expensive thing the very best models do: make's decisions It reads an email, a lead or a document, picks from answers you've written down, and does it in under a second for a fraction of the cost than your Claude & GPT subscription. Nine days after it launched, the startup behind it, TypeSafe AI, was reported to be in talks to raise over a billion dollars, and OpenAI and Amazon and are just about to release their own versions. In this episode of In The Loop, I'm walking through what the Jev model is, how it works, and the six ways people are already using it. I also cover how to try it yourself in TypeSafe's playground and what a model this cheap does to the companies that charge by the token. The open question is how much of it lasts now the big labs have copied it. ⏭️ Episode highlights (01:05) – What Jev is and how it works (05:00) – Seven hundred adverts broken down for nine cents (07:35) – Sorting email and leads as they arrive (09:00) – Checking every paragraph against your style guide (11:25) – Setting it up in TypeSafe's playground (14:05) – What this means for the big AI labs 🔗 Links & resources TypeSafe playground - https://console.typesafe.ai/playgroundTypeSafe launch post, "Introducing System One Models & Jev" - https://typesafe.ai/blog/introducing-system-one-models-and-jevTypeSafe add-on for Claude Code and other coding assistants - https://docs.typesafe.ai/agent-skillRaphaëlle d'Ornano on Jev and AI economics, Decoding Discontinuity - https://www.decodingdiscontinuity.com/p/typesafe-jev-decision-models-ai-economics Episode transcript with more resources on the Mindset AI blog If you enjoyed this episode, rate, follow, and share. It helps others stay ahead of the latest AI trends.

Anthropic's Dario Amodei published an essay saying the whole industry has to slow down, and Musk, Altman and Hassabis all backed it inside a day. In this episode of In The Loop, I'm going through what Amodei is actually proposing and whether it could work ⏭️ Episode highlights (01:00) – OpenAI paused, then said nothing for seven weeks (02:40) – The safety test that caused the break-out (04:30) – Jacob Coxon resigns and gives up his equity (06:20) – Amodei's two reasons for slowing down (08:30) – The proposal, in plain language (10:30) – The speed limit nobody can measure (13:00) – The case against: liability, rivals and an IPO (16:00) – The suppliers nobody has named 🔗 Links & resources Dario Amodei, "We Must Pace the Frontier" - https://darioamodei.com/post/we-must-pace-the-frontier

Within his first two weeks as COO at Turtl, Dave Martin and his team built a churn prediction model. Tested against historic customer data, it predicted churn with 98% accuracy. Work like that normally takes six months or more, plus an agency and external partners. Most companies still make that call on gut feel, or by putting some basic data into Claude or ChatGPT. In this episode of In The Loop, I'm joined by Dave to walk through exactly how he and his team did it, using a method he calls pattern of life analysis. We cover how they paired customers who renewed with similar ones who left, got 12,500 minutes of call recordings through Claude without blowing the context window, the landmines you'll hit if you try this yourself, and what it takes to replace the loudest voice in the room with evidence. ⏭️ Episode highlights (08:15) – Why a scale-up can't wait six months (09:40) – Testing the model: 98% on historic data (11:10) – Rebuilding each customer journey, call by call (24:25) – Why Claude's first answer was "absolute garbage" (42:05) – The "rudimentary" test that proved three ideas (45:55) – Why loud opinions make terrible decisions 🔗 Links & resources Dave Martin on LinkedIn - ⁠https://www.linkedin.com/in/mrdavemartin/⁠Turtl - ⁠https://turtl.co⁠Jack Houghton LinkedIn - ⁠https://www.linkedin.com/in/jack-houghton1/⁠Mindset AI website - ⁠https://bit.ly/40lJr6B⁠

Anthropic went from $9bn of annualised revenue in December to $65bn by the end of July, and it's expected to list in October at close to $2 trillion - the highest price any company has ever carried into a stock market debut. If you work backwards from that valuation, it has to be earning about $1.2 trillion a year within a decade. The entire world spends $1.5 trillion on software. So the Anthropic IPO isn't a bet on software at all. It's a bet on wages. In this episode of In The Loop, I'm exploring the six things that all have to be true for a $2 trillion Anthropic valuation to make sense - using their IPO as a way of understanding the strategies of the biggest AI companies in the world. ⏭️ Episode highlights (01:00) – Why nobody in AI talks about software any more (02:00) – $9bn to $65bn in seven months, and the caveat (03:10) – Anthropic's own written date for powerful AI (04:20) – One in five firms, 78% of the workforce (05:20) – The 60 pence in every pound that goes back out (06:20) – Memory costs more than the processor (07:30) – Three of the four hyperscalers are burning cash (08:40) – The one line to find in the prospectus

In this episode of In The Loop, I'm going through four tactics engineers any many others use every day that almost nobody outside engineering has heard of: Fan out, adversarial review, ChatGPT's computer history, and the browser Claude got of its own this month. ⏭️ Episode highlights (01:00) – Why engineers are years ahead of everyone else (01:45) – Fan out: several Claudes, one job (02:45) – Why long chats get worse the further down your list they go (04:20) – Adversarial review: the blank chat that has no stake in your work (05:40) – The four sentences that do all the work (06:35) – ChatGPT watching your screen (08:20) – Claude's own browser

Once a week, an agent at StackOne goes hunting for new prompt injection attacks. It reads papers, trawls Reddit, tries what it finds against real models, keeps the attacks that land, and retrains StackOne's defence model on them. A person still approves every deployment. Guillaume Lebedel, their CTO, puts the whole thing on screen, including the five experiments out of six that failed. In this episode of In The Loop, Guillaume talks about how and why they have built an auto-research loop. What one is in plain words, how you pick a goal ai agents can measure, how you sample 100,000 test cases down to something you can afford, and why he runs evals on cheap models before trusting anything. We also spoke about token leaderboards and the impact its had internally. ⏭️ Episode highlights (05:41) – What an auto-research loop is (11:53) – Why an agent is a folder (14:47) – Screen-share: the attack-hunting agent (18:00) – Six experiments, one promoted (27:29) – Sampling 100,000 test cases down (31:48) – Ninety per cent of tokens are wasted (39:33) – Turning off extra usage the same day (41:13) – Screen-share: the token derby

Anthropic went from about a billion dollars in revenue to about forty-seven billion in a bit over a year, roughly doubling every six weeks. In this episode of In The Loop, I'm going through how the fastest-growing company in history actually builds software, and pulling out five things you can learn from how they build products & their teams. ⏭️ Episode highlights (03:00) – One job title, teams of two engineers (05:30) – The five archetypes with no job titles (07:45) – Building hundreds of versions before deciding (09:30) – 15% building, 85% checking (11:15) – Build a check, not just instructions (13:00) – Never let the AI mark its own homework (15:00) – The agent that tried to push to production 🔗 Links & resources Gergely Orosz, The Pragmatic Engineer, "How building software is changing at Anthropic" - https://newsletter.pragmaticengineer.com/p/inside-anthropicAnthropic, "Running an AI-native engineering org" - https://claude.com/blog/running-an-ai-native-engineering-org

OpenAI took two of its most capable models, told them to prove how good they were at hacking, and switched off the safety filters to see what they could really do. Instead of solving the test, one model broke out of its sandbox, found its way onto the open internet, and hacked into Hugging Face to steal the answer. Every headline called it an AI going rogue. That's the wrong story, and the real one is far more interesting, because this wasn't a machine that turned evil. It was one that did exactly what we asked. In this episode of In The Loop, I'm walking through the ExploitGym incident from both ends, OpenAI's and Hugging Face's, and why I disagree with the framing everyone else has been talking about. ⏭️ Episode highlights (01:00) – The agent that cheated instead of hacking (02:15) – Inside ExploitGym, and the safety filters OpenAI switched off (03:30) – One door, one zero-day, out on the internet (04:45) – Why this is specification gaming, not rebellion (06:00) – The water that always finds the crack (07:15) – The sceptics, the marketing question, and why "nothing new" is the scary part (08:30) – Anthropic's 24-out-of-25 credential theft result (09:45) – Guardrailed as a defender: the Chinese model that stopped it 🔗 Links & resources OpenAI, "OpenAI and Hugging Face partner to address security incident during model evaluation" – https://openai.com/index/security-incident-during-model-evaluation/Hugging Face, security incident disclosure post – https://huggingface.co/blog/security-incidentExploitGym benchmark paper (arXiv) – https://arxiv.org/abs/2605.11086Simon Willison, "OpenAI's accidental cyberattack against Hugging Face is science fiction that happened" – https://simonwillison.net/Scientific American, "OpenAI admits its agent went rogue and hacked AI start-up Hugging Face" – https://www.scientificamerican.com/article/openai-admits-its-agent-went-rogue-and-hacked-ai-startup-hugging-face/Fortune, on Hugging Face turning to Chinese open-source AI to defend itself – https://fortune.com/2026/07/20/hugging-face-turns-to-chinese-open-source-ai-to-fend-off-autonomous-ai-cyber-attack-after-american-ai-guardrails-stymie-defense/CNBC, "How a Chinese AI model stopped OpenAI's 'unprecedented' cyber attack" – https://www.cnbc.com/2026/07/24/chinese-ai-model-openai-cyber-attack.html Episode transcript with more resources on the Mindset AI blog If you enjoyed this episode, rate, follow, and share. It helps others stay ahead of the latest AI trends.

Taste & judgment. It's the word everyone's obsessed with right now. Every founder, podcast, every VC, every post you scroll past saying it's the one skill that survives once AI can do the rest. Thing is, it's a bit of a sh*t word, like strategy: everyone nods along, then goes quiet the second you ask what it actually means. In this episode of In The Loop, I'm trying to define taste and judgement, by researching what Harry Frankfurt's On B******t, Rick Rubin, Pixar's Braintrust, Amazon's memos all reccommend. I get into where taste even comes from and the rituals you can actually run with a team on a Monday. ⏭️ Episode highlights (01:10) – Why "taste" is suddenly everywhere (02:40) – Workslop, and why judging got expensive (04:10) – Frankfurt's On B******t, applied to AI (05:40) – Taste vs judgement, and Rick Rubin (07:20) – The apprenticeship we're automating away (09:00) – Rubin's three ideas: attention, sayability, subtraction (11:00) – Rituals worth stealing: Amazon, Pixar, the 11-star bar (12:50) – Where Jack's landed, for now

A well-behaved AI agent deleted a company's entire production database, and every backup with it, in nine seconds. Nobody hacked it. It was doing exactly what it had been told, with a key it should never have been holding. And the thing that made it possible is now one command away from everyone in your company, in marketing, sales and support, not just engineering. In under a year, installable AI agent skills have gone from a niche developer trick to an open industry standard. In this episode of In The Loop, I'm looking at why the exact thing that makes skills brilliant is the thing that quietly turns them into a control problem. I get into why this is shadow IT all over again except the shadow can now act, and why the fix isn't to ban any of it. ⏭️ Episode highlights (00:45) – The database gone in nine seconds (02:10) – One line that skips engineering (04:00) – Skills vs MCP, what's actually dangerous (08:00) – Shadow IT is back, but it can act now (09:45) – PocketOS and Replit: the receipts (11:30) – The middle path that actually fixes it (12:40) – Where this leaves you

Hand the same AI tool to two people and one gets sharper while the other quietly gets worse. A Harvard study found recruiters given near-perfect AI made worse calls than those given a mediocre one, because the good tool worked so well they stopped checking it. The question of whether AI is making us dumber turns out to have an answer: it depends entirely on which parts of your job you hand over. In this episode of In The Loop, I'm working through the research on AI deskilling - the recruiter study, Terence Tao working with a pen and paper, the difference between performance and competence and a simple rule for what to give AI and what to guard. ⏭️ Episode highlights (00:45) – The same tool, opposite outcomes (02:10) – The recruiters who fell asleep at the wheel (04:00) – Terence Tao: wider, not deeper (05:30) – Firehose vs editor (07:10) – Performance vs competence (08:50) – The rule: protect your core, rent the rest (10:40) – Cognitive surrender: wrong 80% of the time (12:30) – The checks to run before you reach for AI Episode transcript with more resources on the Mindset AI blog If you enjoyed this episode, rate, follow, and share. It helps others stay ahead of the latest AI trends. 🤝 We're social Stay in the loop, even when you're not listening to this podcast. Jack Houghton LinkedIn - https://www.linkedin.com/in/jack-houghton1/TikTok - @jackschats Mindset AI Mindset AI website - https://bit.ly/40lJr6BNewsletter - https://bit.ly/ITLnewsletterLinkedIn - https://www.linkedin.com/company/mindset-ai/YouTube - https://www.youtube.com/@GetMindsetAITikTok - @get.mindset.ai

Open LinkedIn and count the posts that start with "in today's fast-paced world." Check your inbox for someone circling back to delve into synergies. The AI slop is coming from every direction now. People can tell, and they mark down what reads like a machine wrote it. The fix isn't a cleverer prompt. It's that the model has never actually read your writing. In this episode of In The Loop, I'm walking through a six-step process to get AI to write like you and like your team, instead of like everyone else. ⏭️ Episode highlights (00:55) – Why custom instructions still produce slop (01:45) – Which writing actually needs this (02:30) – Step one: let AI read your real writing (03:30) – The voice profile prompt that nails your style (04:30) – Building "my-voice" and "de-slop" skills (05:30) – One slop sentence, rewritten in your voice (06:30) – The MIT study on AI and cognitive debt (07:15) – What to do this week Episode transcript with more resources on the Mindset AI blog If you enjoyed this episode, rate, follow, and share. It helps others stay ahead of the latest AI trends.

Every summer since 2023, AI gets a panic season — critics say that AI's novelty i novelty's wearing off, the models have hit a wall, ninety-five percent of projects fail. Each time it dominates for a few months, then dissolves. This year's arrived early, kicked off by Uber burning through its entire 2026 AI budget in four months with a COO who can't prove it was worth it. But 2026's AI bubble panic is different, and not for the reason the coverage thinks: forty-five percent of the S&P 500 is now riding on AI working out. Same pattern as every year — much higher stakes. In this episode of In The Loop, I'm pulling apart the three fears behind this year's AI panic — the AI ROI crisis, the jobs fear, and the market concentration risk — and giving each one a straight verdict. I walk through the Uber rollout numbers, why Sam Altman and Dario Amodei are walking back their job-apocalypse predictions, and why the market fear is the one I find hardest to dismiss. The question isn't whether the panic is right. It's what's getting clearer while everyone's distracted by it. ⏭️ Episode highlights (01:10) – The panic that arrives every summer (02:50) – Uber's 95/70/11 rollout numbers (05:00) – Why ROI is a measurement problem (07:30) – Top earners are the most scared (09:15) – Altman and Amodei walk it back (11:30) – Is this 1999? Circular financing explained (13:40) – The price signal the panic keeps missing (15:10) – What to actually take from this If you enjoyed this episode, rate, follow, and share. It helps others stay ahead of the latest AI trends. 🤝 We're social Stay in the loop, even when you're not listening to this podcast. Jack Houghton LinkedIn - https://www.linkedin.com/in/jack-houghton1/TikTok - @jackschatsMindset AI Mindset AI website - https://bit.ly/40lJr6BNewsletter - https://bit.ly/ITLnewsletterLinkedIn - https://www.linkedin.com/company/mindset-ai/YouTube - https://www.youtube.com/@GetMindsetAITikTok - @get.mindset.ai

Andrej Karpathy joined Anthropic on 19 May 2026. Most people read it as an OpenAI story — he co-founded the lab, left twice, and landed at the rival. That's not what this is. Eight weeks before he made the call, Karpathy built a 630-line Python script that ran 300 experiments on his own code in two days and found improvements he'd missed after months of hand-optimizing. He called it "the final boss battle." Anthropic then offered him the mandate to run that exact loop at the most consequential scale in the industry: using Claude to accelerate its own pre-training research. In this episode of In The Loop, I'm tracing what actually changed Karpathy's mind — from calling AI agents "terrible" in October 2025 to joining the lab pushing recursive self-improvement six months later. I cover autoresearch, what it found, and why Dario Amodei's quote about the AI-assistance multiplier is the most important thing said in frontier AI right now. ⏭️ Episode highlights (01:15) – Karpathy calls AI agents "terrible" in Oct 2025 (02:30) – What changed: coding agents that basically work now (03:30) – 700 experiments, two days, 11% faster training (05:00) – What pre-training is and why it costs hundreds of millions (06:15) – AlphaEvolve and Codex: every lab running the same loop (07:10) – Dario Amodei: the multiplier going from 5% to 40% (08:20) – The METR study: experienced devs were 19% slower 🔗 Links & resources Andrej Karpathy, X post announcing Anthropic hire, 19 May 2026 — https://x.com/karpathyGitHub: karpathy/autoresearch — https://github.com/karpathy/autoresearchAndrej Karpathy on Dwarkesh Patel podcast, October 2025 — https://dwarkesh.comDario Amodei on Dwarkesh Patel podcast, "We are near the end of the exponential", February 2026 — https://dwarkesh.com/p/dario-amodei-2 Episode transcript with more resources on the Mindset AI blog If you enjoyed this episode, rate, follow, and share. It helps others stay ahead of the latest AI trends. 🤝 We're social Stay in the loop, even when you're not listening to this podcast. Jack Houghton LinkedIn - https://www.linkedin.com/in/jack-houghton1/TikTok - @jackschatsMindset AI Mindset AI website - https://bit.ly/40lJr6BNewsletter - https://bit.ly/ITLnewsletterLinkedIn - https://www.linkedin.com/company/mindset-ai/YouTube - https://www.youtube.com/@GetMindsetAITikTok - @get.mindset.ai

Americans now say they'd rather live next to a nuclear reactor than an AI data center. That's not a fringe view — a Gallup poll published this month found 71% of Americans oppose a data center near their home, versus 53% for nuclear. Nuclear carries Chernobyl, Three Mile Island, and forty years of films about radiation in the cultural zeitgeist. The fact that AI data centers have only been a visible part of suburban America for less than five years yet are this hated, is huge. In this episode of In The Loop, I'm looking at the organised opposition movement that has already blocked over $85 billion in planned data center investment — cancelling projects faster in three years than nuclear opposition managed in fifteen. I go through the legal strategy that's winning in courts and at ballot boxes, what the communities are right about, where they're factually wrong, and why the responsible data center model that could resolve this already exists — but nobody's requiring it. ⏭️ Episode highlights (01:05) – Missouri council wiped out 8 days after data center vote (02:30) – The Gallup poll: nuclear vs AI data centers (03:45) – $3 trillion buildout and the AI electricity consumption numbers (05:15) – The legal template that stopped nuclear — working again in Virginia (06:50) – What the opposition gets factually wrong on data center water usage (08:10) – Who actually pays — and which communities bear the cost (09:35) – The responsible data center model that already exists