The Derby Mill Series: Pushing AI to the Limit

Many CEOs say that deciding who to hire is the most important decision a company makes. Many AI experts say that temporal difference learning is one of the most important concepts in AI. In this episode, The Derby Mill team draws the link between the two.Although the hiring decision is so critical, traditional recruitment is notoriously inefficient and susceptible to human bias. How can AI recruiting agents help companies source and evaluate talent? How much of the hiring process can - and should - be handed over to AI? The Derby Mill team joins Metaview co-founders Siadhal Magos and Shahriar Tajbakhsh to tackle these questions. AI already has affected hiring, with annual applications per recruiter increasing by approximately 200% between 2022 and 2025. Recruiters spend their days processing massive application queues, writing outreach, scheduling screens, and shuttling information between systems. The average time to fill a role is nearly 45 days, and nearly 50% of hiring decisions are later regretted.To help client companies navigate the slop, Metaview recently launched fillmore, an agentic AI designed to act as an autonomous recruiting coworker. Instead of recruiters spending weeks sourcing leads and drafting job descriptions, fillmore ingests intake conversations, drafts role requirements, and surfaces targeted candidate matches for human review. Metaview says thousands of companies are already deploying fillmore, conducting a growing number of candidate interactions and shortening time-to-hire by over 75%.In this episode, the discussion begins with an examination of the recruiting funnel and where AI delivers real leverage. Sendhil Mullainathan has thought deeply about bias in the hiring decision. Over twenty years ago he published what became a landmark study of racial bias in hiring. The panel discusses how AIs can reduce human bias.Later, they investigate the future of AI-enabled hiring. Rich Sutton notes that “It’s an interesting learning problem... an exact case where you have guesses that are trained by later guesses,” linking the interview process directly to temporal difference learning.Applied AI scientist Niamh Gavin explores how organizations can track counterfactual outcomes for rejected candidates, and whether AI can ever truly replace high-stakes, face-to-face evaluations. Finally, Ajay and Sendhil discuss how agentic AI could fundamentally reshape organizations—moving hiring away from individuals toward the onboarding of entire teams or departments.What’s certain is that hiring and recruiting is set to change as agentic AI streamlines its processes - and Metaview is at the forefront of this new corporate landscape.FEATURINGSiadhal Magos, co-founder and chief executive officer, MetaviewShahriar Tajbakhsh, co-founder and chief technology officer, MetaviewAjay Agrawal, co-founder and partner, Intrepid Growth PartnersSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant Niamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaLINKSMetaview’s value proposition explained. Metaview’s website. Metaview’s recent Series C fund raise announced here with participation from Intrepid GP.Mentioned in this ep: Sendhil’s research on hiring bias in American Economic Review. PDF of the working paper. Rich Sutton’s bitter lesson essay.Subscribe to The Derby Mill Series at our Substack (main site) or on YouTube, Spotify or Apple Podcasts.Derby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau.NUGGETSRethinking the Job DescriptionCan AI help a company better understand the role it’s filling?Everyone’s RecruitableRather than relying on applicants, Rich Sutton and Siadhal Magos believe AI-enabled hiring will recruit from the entire global populationNo More Hiring IndividualsAjay and Sendhil suggest AI-enabled hiring will lead companies to recruit entire teams rather than “onesies”Recruiting as AI Learning ProblemRich Sutton observes that recruiting presents a unique learning problem wherein guesses are trained by later guessesDISCUSSION POINTS00:00 Cold Open00:30 Human Bias & fillmore: The AI Recruiter05:15 How AI Speeds Up Job Posting & Finding Candidates10:45 The Tech Behind AI Sourcing & Real-Time Matching19:35 Human Assessment vs. Algorithmic Sourcing31:29 Predicting the Future of AI Recruitment46:32 Rethinking Company Workflows: Onesies vs. Team Hires53:19 How Metaview Can Encourage Exploration01:02:18 Final Thoughts01:07:29 Wrap-upDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

Earlier this month, JPMorgan CEO Jamie Dimon remarked, “I wrote about cyber this year. I said, and I meant it, it’s our biggest risk. And that was a couple weeks before Mythos. And now it is our biggest risk by far.” Then last week, The Wall Street Journal ran the headline “AI Just Found a Bug That Puts up to 70% of the World’s Websites at Risk; Artificial-intelligence tools helped find one of the most severe online bugs ever. Cyber experts are racing to patch it before hackers use AI to exploit it.” Cybersecurity researchers, the accompanying story noted, had issued an alert about two AI-uncovered vulnerabilities in PHP (PHP: Hypertext Preprocessor), a programming language that powers most of the web, including WordPress and Wikipedia. The same day, The Information reported a totally separate cyber attack: “A hacking group called CyberCartel used a web of AI agents powered by models from companies including Anthropic, SpaceX and Chinese lab Moonshot AI to automate cyberattacks against more than 400 Latin American financial targets, according to an investigation by the cybersecurity startup Straiker.” Two days later, The Atlantic ran the headline “Welcome to the Valley of Pain: An explosion of AI-powered hacks will reshape life online.”All this was forecast at Intrepid’s June annual general meeting in London. The Derby Mill team and Gianni Cuozzo, the CEO and founder of Rome-based Exein S.p.A., an Intrepid portfolio company that produces cybersecurity software for AI-enabled physical connected devices, anticipated the coming surge in cyber attacks.Recent advances in AI lower two costs that increase the likelihood of cyber attacks: 1) they drastically lower the cost of finding vulnerabilities in code, 2) they severely lower the cost of exploiting those vulnerabilities. While advances in AI also lower the costs of cyberdefence, the impact is not symmetric. The reason AIs advantage attackers more than defenders is due to their unpredictable error profiles, which are less problematic for attackers.Sendhil Mullainathan summarized, “Both the discovery and the exploit [of the coding flaws] are now about to get orders of magnitude much bigger. So if you think of a tail risk for the entire economy, this type of thing is sitting on that. You could imagine an attack on the New York Stock Exchange that caused chaos or rewrote trades even just for 5 minutes…. it’s arguably the biggest near-term threat that is not really described because you forget how much of the economic system, the medical system, everything relies on technology.”PARTICIPANTSGianni Cuozzo, founder and CEO, Exein S.p.A.Ajay Agrawal, co-founder and partner, Intrepid Growth PartnersSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsLINKSFind Exein on the web. Context on their $270M raise, led by Headline with participation from Intrepid Growth Partners and others, at a $1.7B valuation for building the security layer for physical AI. Report on the round from the Financial Times. Intrepid’s report. More PHP vulnerability context from the BBC, the SANS Institute and the New York Times.Hugging Face incident context. MIT Technology Review. OpenAI’s take. Op-Ed from The Guardian. Commentary from WSJ.Derby Mill’s previous conversation with Gianni Cuozzo, which predicted incidents like Hugging Face.Subscribe to The Derby Mill Series at our Substack (main site) or on YouTube, Spotify or Apple PodcastsDerby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau.DISCUSSION POINTS00:00 Cold open00:40 Exein Origins06:41 Agnostic Architecture08:03 Product Ecosystem09:22 Machine-Based Ransomware11:44 AI Agent Communications14:03 Escalating Tail Risks18:44 On-Device Embedded Security21:39 Embedded System Vulnerabilities24:02 Audience Q&A25:50 Market Demand Dynamics27:12 Azure Container Incident32:00 White Hat Hacker Bounties33:08 Asymmetry in LLM Capabilities36:06 Real-Time Processing Constraints39:34 LLM-Generated Code Vulnerabilities41:40 Long-Term HorizonDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

The nonprofit AI-progress assessment organization, METR (Model Evaluation and Threat Research) released a graph recently that has become “an industrywide obsession,” according to the New York Times. But are there better ways to track AI progress? Here, the Derby Mill panel weighs in.METR uses a task-completion time horizon to assess the progress of AI frontier systems. The “time horizon” is the duration that it would take a human expert to complete a set of diverse software development tasks.METR’s 50% time horizon, then, is the “duration at which an agent is predicted to succeed” at a given software development task half the time.According to METR, frontier models over the last year have increased their task-completion horizon by about an order of magnitude, from roughly 1.5 hours to about 15 hours (see graph).In this episode, Derby Mill host Ajay Agrawal begins by challenging Rich Sutton, Suzanne Gildert and Niamh Gavin: Should we expect another order-of-magnitude increase over the next 12 months?The discussion quickly turns into a debate over the authenticity of METR’s measurement. Does the METR graph, and its remarkable improvement, reflect an improvement in AI model intelligence?“I think this stuff is totally bogus,” says reinforcement learning pioneer Rich Sutton. “You have to realize how easy it is to cook these numbers.”Rich’s criticism prompts Ajay to challenge the panel: Is there a way to measure AI’s progress that is better than METR’s method? Niamh discusses her role in a fascinating initiative to assess AI models by how well they win elections in simulated environments. And what merit is there in Suzanne’s suggestion, to assess progress based on aggregate numbers of job postings for software developers?FEATURINGAjay Agrawal, co-founder and partner, Intrepid Growth PartnersNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsSuzanne Gildert, CEO, Nirvanic Consciousness Technologies, quantum physicist, co-founder of Sanctuary AI and KindredRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaLINKSMETR’s task-completion time horizon model explained, including the various versions of the time horizon graph at 50% and 80% success rates. Scroll down for a great FAQ explaining the effort’s various points.METR CEO Beth Barnes in conversation on 80,000 hours.The NY Times article that observed how METR’s graph has become “an industrywide obsession.”Derby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau.DISCUSSION POINTS00:00 Cold open00:27 Task Horizon Graph Explained06:01 Low Context vs. High Context Tasks10:10 Critiquing Model Evaluation Metrics15:53 Testing AI Capabilities20:08 Alternative Measures27:07 Do We Need Quantitative Benchmarks?DISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

Atinary Technologies, based in Lausanne and Silicon Valley, is pioneering autonomous scientific experimentation with AI-driven robotic research labs. In this episode, our season finale, the Derby Mill panel welcomes Hermann Tribukait, co-founder and CEO of Atinary, whose vision is to use the company’s technology to exponentially accelerate the discovery of breakthrough molecules.Named after the Spanish verb “atinar” — to hit a target — Atinary’s product is SDLabs, short for “Self-Driving Labs Platform,” a code-free agentic AI solution that enables scientists to optimize their experimentation workflows without coding or machine-learning expertise. The platform has been deployed across pharma, chemicals, and materials R&D.Earlier this year, in a significant step from the digital to the physical, Atinary opened its first Self-Driving Lab in Boston, where it runs experiments autonomously 24/7. Reported efficiencies and scaling are impressive. Atinary says it can generate in a week the data that would take a grad student an entire Ph.D. For one client, Atinary’s self-driving experimentation cut the use of an expensive catalyst by 30x, reducing cost by 97% and reaction time by 50%.Is autonomous scientific discovery realistic? And where might the technology go at the limit? “One place where all this leads,” observes Sendhil, “Chemistry will get one step closer to a solved field—and when chemistry gets close to a solved field, boy does that look completely different.”PARTICIPANTSHermann Tribukait, co-founder, CEO and chair, Atinary Technologies Inc.Ajay Agrawal, co-founder and partner, Intrepid Growth PartnersSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsLINKSAtinary websiteLoïc Roch, Atinary CTO and co-founder, explains the technologyAtinary launches its first self-driving labSubscribe to The Derby Mill Series at our Substack (main site) or on YouTube, Spotify or Apple PodcastsDerby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau.DISCUSSION POINTS00:00 Cold open00:52 Explaining Atinary03:40 Sample objectives05:54 DMTA and learn07:22 Efficiencies09:29 Scaling challenge12:39 Why run experiments?15:12 A solved field16:30 Chemistry and RL17:47 GenAI lessons18:44 Football fields20:22 Chemistry as biology21:29 Formulation challenge23:15 Combinatorial search26:52 Programmable biology28:57 OFAT 30:44 Lightning round33:12 Eroom’s lawDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

In our latest episode, Intrepid’s Derby Mill podcast features the CEO and co-founder of UK-based PhysicsX, Jacomo Corbo, for a special in-person episode filmed at our Annual General Meeting in London. The interview, conducted by host Ajay Agrawal and panel members Sendhil Mullainathan and Niamh Gavin, comes just weeks after PhysicsX announced an oversubscribed $300 million Series C investment round that valued the company at $2.4 billion, which included Intrepid participation.The PhysicsX origin story includes Corbo and his co-founder, Robin Tuluie, working on the Formula 1 racing circuit for the Renault team that won back-to-back world championships in 2005 and 2006. (Corbo took a year off from studying computer science for his Ph.D. at Harvard after the Renault team noticed his game-theory based research and asked him to apply it to F1 racing environments, eventually becoming Renault’s chief race strategist.)Corbo and Tulule started PhysicsX in 2019 and now say that their machine-learning foundation models can accelerate simulation for industrial engineering by between 10,000 to 100,000 times. That means more efficient product design cycles, Corbo says, delivered under faster timelines, which in turn translates into better products. Based in London and New York with more than 300 employees, PhysicsX counts among its clients leading organizations in aerospace & defense, automotive, semiconductors, materials, and energy & renewablesOn the agenda in today’s discussion: How do the PhysicsX AI-enabled emulation models differ from conventional simulation? What capabilities does better simulation unlock for today’s industrial engineers? And how will tomorrow’s products differ as a result? In our first-ever in-person episode filmed in London, the Derby Mill tackles it all.PARTICIPANTSJacomo Corbo, co-founder and CEO, PhysicsXAjay Agrawal, co-founder and partner, Intrepid Growth PartnersSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsLINKSPhysicsX websitePhysicsX press release for its last investment round, in which Intrepid participatedHarvard School of Engineering alumni profile on Jacomo CorboSubscribe to The Derby Mill Series at our Substack (main site) or on YouTube, Spotify or Apple PodcastsDerby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau.DISCUSSION POINTS00:00 Cold open01:10 PhysicsX described03:10 AI Numerical Inference04:00 Governing Physics Equations05:57 Engineering Intuition Limitations08:07 Optimization Computational Barriers10:01 Democratizing Engineering Simulation11:05 Deep Learning Models11:26 Empirical Data Training12:28 Industrial Customer Segments14:41 Approximating Ground Truth15:55 Expanding Design Space17:01 Dark Factory Vision18:56 Key Breakthrough Requirements21:05 Training Data Scarcity22:10 Reinforcement Learning Approach24:59 End-of-One Customization26:27 Influencing Physical World28:31 Scaling Pre-trained Models29:21 Model Attention Mechanisms30:35 High Dimensional Physics32:41 Reducing Exploration Costs35:12 Manufacturing Cost Curves36:52 Sim-to-Real Performance38:15 Improving Causal Inference38:40 Quantum Computing Integration40:01 Innovativeness and NoveltyDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

UK-based Relay is a fast-growing delivery company whose mission is to lower the cost of last-mile fulfillment. CEO and co-founder Jonathan Jenssen say his team is reinventing networks for modern e-commerce. Relay covers about 23% of the UK, has a 4% share in operating postal codes and includes on its customer list such brands as Vinted, Temu, Shein, TikTok Shop, ASOS and Next.One big challenge for Relay is that the last mile amounts to ~70% of total delivery cost. Also confounding cost decreases is the fact that the parking-to-doorstep journey amounts to ~80% of last-mile time. That all means that about half of the total cost of package delivery is in the last 100 metres of the delivery.To decrease those costs, Jenssen’s team is harnessing the power of AI to make the last few hundred metres of the delivery more efficient. Gone are the days of hub and spoke routing and half-empty delivery trucks. Algorithms provide couriers with next-best-actions, in the form of turn-by-turn driving and parking directions, tips for building entry, and doorstep handover. They also use dynamic per-route pricing and live supply/demand matching, dynamic routing and an asset-light pit stop network of delivery depots that currently includes 10,000 local shops.The Derby Mill team explores where Relay’s efficiency improvements may go at the limit of AI’s capabilities. Where could robots fit in? What if Relay had access to more consumer information? Could that allow the company to actually predict customer requests? And what does all this have to do with something called “the beer game,” an MIT Sloan business case? These questions and more are answered in episode 31 of The Derby Mill Series. HOSTS AND PANELLISTSJonathan Jenssen, co-founder and CEO, RelayAjay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsSuzanne Gildert, CEO, Nirvanic Consciousness Technologies, quantum physicist, co-founder of Sanctuary AI and KindredLINKSRelay websiteMIT Sloan beer distribution gameSubscribe to The Derby Mill Series at our Substack (main site) or on YouTube, Spotify or Apple PodcastsDerby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau.DISCUSSION POINTS00:00 Cold open00:43 Introductions01:35 Explaining Relay05:02 Relay video06:28 Clarifying questions: Rich09:50 Optimization parameters11:55 Clarifying question: Suzanne16:22 Sendhil Q: Tech unlock22:50 Hubs & spokes26:40 Niamh: Pricing routes29:33 Reward signals32:23 Suzanne on robots36:30 Coordination elements39:03 Beer game44:02 Information bottlenecks47:45 Agentic info passing49:30 JJ responds51:40 Lightning round53:33 Agentic AI agency55:51 Supply chains57:20 Wrap upDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

How are companies using AI to improve the healthcare journey for consumers? And how will things evolve at the limit, where wearable technology integrates with electronic medical records and medical agentic AI analyzes everything on the fly to issue recommendations and improve outcomes? In other words, what are the implications of everyone becoming equipped with their own superhuman medical agent? Seen one way, a superhuman medical agent is what League is aiming to develop. Founded in 2014 in Toronto, League operates an AI-powered healthcare platform for insurance companies and other payers or providers. The platform allows users to manage health benefits as well as their own healthcare. As of early 2026, the company has more than 60 million contracted users and more than a billion consumer interactions annually. Customers include Highmark Health, Manulife, Shoppers Drug Mart, and Medibank. Last year, League gave 100 million healthcare recommendations, of which people clicked through or followed through about 65% of the time.   In this episode, the Derby Mill team explores where League’s technology might go at the limit with League founder and CEO Mike Serbinis, who aims to expand the company’s capabilities to create a 24/7 infinite care platform—basically, an all-knowing doctor who knows as much as possible about you, your healthcare data, and the science of medicine. It’s a fascinating conversation that sees Sendhil observing how little the medical profession currently knows about optimizing health, or patients themselves, and Ajay challenging Rich to decide whether superhuman AI doctors would be better than having a doctor as a spouse today. (Rich says yes.)   More on our guest: A serial entrepreneur, Michael Serbinis has also founded and run Kobo, the digital reading company, cloud storage pioneer, DocSpace, and Critical Path, a messaging service. Serbinis is chair of the board for the Perimeter Institute and on the board at the Vector Institute for Artificial Intelligence. He sits on Canada's AI Strategy Task Force.HOSTS AND PANELLISTSMike Serbinis, founder and CEO, LeagueAjay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsSuzanne Gildert, CEO, Nirvanic Consciousness Technologies, quantum physicist, co-founder of Sanctuary AI and KindredLINKS    League website. League explainer videoSubscribe to The Derby Mill Series at our Substack (main site) or on YouTube, Spotify or Apple Podcasts Derby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau. DISCUSSION POINTS00:00 Cold open00:38 Introductions01:19 League explainer01:55 Serbinis explains League05:28 League’s AI today08:09 More AI: Content recommendations12:00 League’s data inputs13:55 AB testing19:10 Healthcare archetypes22:15 Max health v. max engagement25:30 Business model27:00 At the limit: Rich 29:00 Married to a doctor29:42 At the limit: Sendhil31:10 Healthcare becomes you33:37 ROI is huge38:06 Suzanne: Robot paramedics42:53 Serbinis responds46:20 Infinite care team49:16 Sendhil’s wrap upDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

A response from the Derby Mill team of AI experts to Anthropic co-founder Jack Clark’s recent Substack essay, “AI systems are about to start building themselves.” In the essay, Clark predicts that “no-human-involved AI R&D - an AI system powerful enough that it could plausibly autonomously build its own successor - happens by the end of 2028.” Clark notes: “If that happens, we will cross a Rubicon into a nearly-impossible-to-forecast future."With that in mind, Ajay tees up a discussion among Rich, Niamh, Sendhil and Suzanne. Is Jack Clark correct that we’re heading toward a future of self-replicating, self-optimizing AI? What are the implications? Does that actually lead to the singularity? And should we, as Rich Sutton says, treat AI the way a parent treats a child, recognizing that mistakes are inevitable? “There’s some spookiness being implied here,” Mullainathan says, “about it coming alive and taking over a whole set of decisions we didn’t intend to cede it… and if that’s the case, that freaks me out, too.” “The thing that they’re scared about…” says Niamh Gavin. “It’s the fact that you have, in essence, an irreversible positive feedback loop of successive self-improvement cycles, that accelerate it toward what they call a singularity, whereby artificial intelligence exceeds human intelligence and control.”Are the concerns outlined by Jack Clark warranted? Or are the fears about a harmful singularity overblown? The Derby Mill experts provide their views in our latest episode. Finally, another fascinating thread in the episode comes from Suzanne Gildert. “It all comes down to the reward function,” says Gildert. “So there’s this thing that [AI is] still limited by us telling them what to do, because we’re the ones who want something… But that’s eventually going to break because the way the whole reward function in ML works at the moment — it’s all based on our economy… It’s based on people exchanging money for goods and services. All that’s going to break if people can’t work anymore... So that’s why I think we really have to understand what the reward function should be, because the way we’re implementing it now is not going to work beyond the point where [AI] can do all human labour for us.” HOSTS AND PANELLISTSAjay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsSuzanne Gildert, CEO, Nirvanic Consciousness Technologies, quantum physicist, co-founder of Sanctuary AI and KindredLINKS    The Jack Clark Substack post that triggered the discussion. Jack Clark is on X @jackclarkSFSubscribe to The Derby Mill Series at our Substack (main site) or on YouTube, Spotify or Apple Podcasts. We post highlights from the show on YouTube.  Derby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau. DISCUSSION POINTS00:00 Cold open01:06 Jack Clark post02:56 Rich Sutton reaction08:37 Sendhil’s reaction16:22 Suzanne’s reaction18:14 Niamh’s reaction23:48 Wrap upDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

Not just an episode about tax law! In fact, the Derby Mill team’s conversation with Benjamin Alarie, the co-founder and CEO behind Blue J, the fast-growing tax research platform, explores numerous fascinating threads. How will AI change the practice of law and the litigation process? Will trying a case in the justice system be necessary when AI can predict with high accuracy the likely outcome? In other words, will AI make litigation obsolete?Separately from the discussion about the legal profession, in a section that will be useful for any entrepreneur seeking to tune foundation models to provide the most current results on fast-changing bodies of knowledge, Alarie describes the techniques that Blue J uses to ensure the platform’s responses reflect the latest regulations. Also useful for entrepreneurs is the discussion on whether specifically tuned, highly specialized models will be more useful to humans than general-purpose LLMs that can provide guidance on a wide variety of topics. ABOUT BLUE J: Founded in Toronto in 2015, Blue J aims to make the law more transparent and accessible for tax practitioners. In early 2026 the company counted 5,000 businesses in the U.S., United Kingdom and Canada as clients who count on it to provide them with answers on numerous different bodies of tax law. (Intrepid is an investor.) A 2025 Series D investment round valued the company at more than $300 million USD, according to the Globe and Mail. The platform is able to predict litigation results with greater than 90% accuracy. “Blue J is now the best way to do tax research, flat out, across any other possible technology being deployed today,” Alarie says, arguing that Blue J is better able to provide guidance based on the current state of fast-changing tax law because it licenses proprietary up-to-the-minute regulatory information from taxation organizations in the jurisdictions it services. Just 0.085% of Blue J’s responses are rated thumbs down by clients and the net promoter score is over 80, with both stats improving as Alarie’s team optimizes the platform. ABOUT GUEST BENJAMIN ALARIE: In addition to being the co-founder and CEO of Blue J, Benjamin Alarie is full professor and holds the Osler Chair in Business Law at the University of Toronto Faculty of Law. He earned a B.A. at Wilfrid Laurier University, an M.A. and J.D. at the University of Toronto, and an LL.M. at Yale Law School. In 2004 he joined the University of Toronto after clerking at the Supreme Court of Canada. His research focuses on tax law, judicial decision making, and artificial intelligence. He has published extensively and is responsible for coining the concept of the "legal singularity" in 2016.HOSTSAjay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsSuzanne Gildert, CEO, Nirvanic Consciousness Technologies, quantum physicist, co-founder of Sanctuary AI and KindredLINKS    Link to Ben’s books: The Legal Singularity. Superjustice. Alarie’s co-authored article: Legal Order in the Age of AI AgentsWebpage for Blue J. Media coverage on Blue J’s 2025 Series D investment round from BetaKit and the Globe and Mail. Another podcast featuring Alarie: The Startup CEO Show. Subscribe to The Derby Mill Series at our Substack (main site) or on YouTube, Spotify or Apple Podcasts Derby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau. DISCUSSION POINTS00:00 Cold open00:48 Blue J and tax law02:20 How Blue J works06:33 Blue J vs. major LLMs12:30 Managing liability15:21 Blue J value prop17:15 The Blue J story19:36 Blue J operating mechanics23:02 General purpose vs. special purpose28:42 Will AI make litigation obsolete?35:10 Rich: Code is law40:55 Ben on code is law45:12 Legal singularity48:26 Faster settlements?51:33 Trend toward legal complexity55:55 Law as learning machine56:51 Niamh reacts57:54 Rich reacts59:50 Suzanne reacts1:00:57 Ben reactsDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

When the right answer is a matter of life and death, how do you ensure that your AI systems avoid fatal mistakes? Signal 1 is a software platform for healthcare that functions as a control panel for deployed AI systems in hospitals. Rather than building individual models, Signal 1 focuses on improving the safety, observability and governability of pre-existing AI models used throughout the patient experience. The platform enables health systems to track real-world performance, detect drift and risk, enforce approval workflows, and tie AI predictions to improved clinical and operational outcomes.Tomi Poutanen is Signal 1’s CEO and co-founder. His earlier start-up, Layer 6, employed artificial intelligence to provide financial services companies with predictive analytics, and was acquired by TD Bank, where Tomi served as chief AI officer. In this episode, Tomi works with the Derby Mill team to discuss the future of healthcare, whether machine learning could make hospitals obsolete, and how to improve the management of systems that include numerous different AI agents working together.  GUESTS AND HOSTSTomi Poutanen, CEO, Signal 1Ajay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist and CEO, Emergent PlatformsSuzanne Gildert, CEO, Nirvanic Consciousness Technologies, quantum physicist, co-founder of Sanctuary AI and KindredLINKS    Signal 1 websiteMedia on Signal 1 from the Globe and Mail, Betakit and the University of TorontoDerby Mill series website. Derby Mill is created by the team at Intrepid Growth Partners and produced by Ghost Bureau.Be sure to catch every episode of The Derby Mill Series by subscribing on the following platforms: YouTube // Spotify // Apple Podcasts // SubstackDISCUSSION POINTS00:00 Cold open00:44 Ajay’s tee up01:22 About Signal 104:11 Tomi’s Signal 1 explainer06:25 Mechanics of Signal 107:10 Tomi’s first-order question08:45 Signal 1’s core value prop09:29 Patient experience perspective12:27 Taking MD out of the loop13:56 Why Signal 1 focuses on AI18:20 Evaluating AI systems22:46 Rows23:13 Ajay’s segue24:42 Part 2: Signal 1 at the Limit25:12 Niamh’s sci-fi hospital vision29:04 Rich questions the premise31:25 Sendhil: Division of labor for AI34:50 Tomi: Niamh’s vision is real38:22 Tomi: Competitive healthcare43:55 Sendhil: Virtues of randomness49:44 Niamh: Multiple agent drama52:54 Tomi’s final remarksDISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com