Leadership for the Physical AI Age

About Leadership for the Physical AI Age

A podcast by Tryfecta Capital exploring the intersection of AI, Robotics, and Leadership. We deconstruct the business of Physical AI: how to invest in it, how to build it, and how to lead through it. Featuring interviews with Industrialists, deep tech founders, VC insights, and market analysis on the future of automation and embodied intelligence.

Publisher

Titto Thomas

Category

business · technology

Language

en

Explicit

No

First episode

2 Apr 2026

Latest episode

28 Sept 2026

Latest episodes

8 of 16 episodes have a transcript.

1. 

   28 Sept 2026

   EP 16: Teaching Robots to Understand Risk

   What does it take to teach autonomous physical systems how to quantify and respect human risk? In this episode of Leadership for the Physical AI Age, host Titto Thomas sits down with Sarah Huth—former Australian Defence Force (ADF) intelligence specialist turned heavy-industry safety and risk practitioner. Drawing from her background in defense intelligence and frontline experience across Australian mining, oil, and gas, Sarah breaks down why workplace safety has remained largely reactive and weighed down by administrative overhead. Together, Titto and Sarah explore the shift from static checklists to proactive risk intelligence: how to train AI agents to bridge the "say-do gap," unify isolated data across operations and maintenance (CAMS), mitigate human cognitive bias and normalization of deviance, and apply Vision Language Models (VLMs) directly to field safety without crossing into invasive surveillance. Key Discussion Highlights Threats vs. Hazards: Applying enterprise risk management principles and battle-space environmental control from the military to heavy industrial worksites. The "Say-Do Gap": Why written safety procedures fail on the ground due to human fatigue, distraction, peer pressure, and procedural drift. Unifying Siloed Systems: Connecting computerized asset management systems (CAMS), daily operations logs, and safety reports into a single, cohesive organism. Stabilizing Risk Tolerance: How AI can standardize Risk Analysis Matrices (RAM), calibrate frontline hazard inputs, and surface weak signals before incidents happen. Industry 5.0 vs. 6.0: The practical realities of keeping humans in the loop alongside autonomous mobile machinery and clustered worksites. Vision Language Models (VLMs) & Governance: Using computer vision for sentiment mapping, situational awareness, and automated compliance tasks—such as performing chemical inventory stocktakes directly from a photo. Safety vs. Privacy: Balancing on-site monitoring and vehicle cameras with worker trust while maintaining the ultimate mission: ensuring everyone returns home safe. Timestamps – Introduction: Welcome Sarah Huth – Transitioning from ADF intelligence into heavy-industry WHS – Security threats vs. safety hazards: The shared foundation of risk – Lessons from overseas tours: Operational perspective and defense trickle-down – The human factor: Cognitive bias and safety rules "written in blood" – Moving from reactive reporting to actionable intelligence across CAMS and operations – Inductive vs. deductive analysis in fast-moving environments – Normalization of deviance and procedural compliance on site – Augmenting the Risk Analysis Matrix (RAM) using AI – Viewing the enterprise as a single data organism – Practical AI: Uncovering low-cost solutions to high-cost problems – Why operational risk is the foundation of physical AI – Closing the "say-do gap" and streamlining the hierarchy of controls – Industry 5.0 vs. 6.0: Collaborative robotics vs. remote autonomy – Worksite command-and-control: Structuring uncontrolled environments – Alleviating administrative double-handling for safety professionals – Visual intelligence: VLMs, fatigue detection, and situational mapping – Navigating the privacy friction of on-site monitoring systems – The vision: Shifting safety leaders from paperwork to frontline coaching – Automating governance: Instant chemical stocktakes via photos and dynamic SDS - Closing thoughts and wrap-up Connect & Resources Host: Connect with Titto Thomas on LinkedIn (https://www.google.com/search?q=https://www.linkedin.com/in/titto-t-b2387319/&utm_source=gemini) Explore Deep Tech & Physical AI: Tryfecta Capital (https://tryfecta.biz/?utm_source=gemini) Subscribe: Follow Leadership for the Physical AI Age on Apple Podcasts, Spotify, and your favorite listening platform so you never miss an episode.

2. 

   14 Sept 2026

   Ep 15: Beyond the Rules: Who Governs Physical AI?

   American farmers hiring Ukrainian hackers to break into tractor software wasn’t just a right-to-repair battle—it was a preview of the defining debate of our era: Who governs physical AI? When every piece of hardware is packed with proprietary sensors, tethered to subscriptions, and generating critical telemetry, traditional concepts of property and liability break down. In this episode of Leadership in the Age of Physical AI, host Titto Thomas and co-host Nick Shelton welcome back futurist and director Melissa Clark-Reynolds ONZM to explore why physical AI will never have a single, unified rulebook—and why real productivity gains won’t come from chatbots writing emails, but from intelligent machines operating in the physical world. In this episode, we discuss: The Ukrainian Tractor Hack: Closed-loop software, third-party parts, and why right-to-repair is the frontline of physical AI autonomy. The Patchwork Regulatory Reality: Why physical AI won't be governed by an "AI Act," but by an ad-hoc mosaic of workplace safety, product liability, road rules, and maritime codes. The History of Job Panic: What the 1906 New York manure export economy and the icemen's union teach us about modern AI displacement—and what the US WARN database actually reveals about tech layoffs. Capability Multipliers vs. Screen AI: Why generative screen tools have left macroeconomic GDP flat, while physical robotics (warehouse automation and robotic weeding) drive real structural productivity. Who Owns the Digital Commons? From Danish facial property rights to Australian seven-metre subsurface land laws—who owns the data generated by an autonomous farm, vehicle, or ocean drone? World Models & Digital Twins: Adapting gaming engine principles from Weta Workshop into low-cost 3D farm simulations for regenerative agriculture, crop planning, and runoff protection. Liability & The 92% Safety Paradox: Why Swiss Re found autonomous driving dramatically safer than human operation, yet human psychology still resists clinical, sterile autonomous design. Off-Planet AgTech & Terraforming: Why solving autonomous agricultural intelligence on Earth is the non-negotiable foundation for sustaining life on the Moon and Mars. What’s Next in Smart Ag: Climate-resilient autonomous greenhouses, in-field mobile cow milking, and drone swarms for pasture management. Timestamps: 00:00 – The Ukrainian tractor hack & right to repair 02:23 – Patchwork governance: Road rules, safety standards, and Tesla in NZ 04:39 – Why regulation lags behind commercial deployment 06:50 – 1906 manure exports, refrigeration bans, and historical tech panic 09:12 – AI as a capability multiplier: Why physical AI drives true GDP 10:33 – Automating the miserable work: Robotic weeding & regenerative soil 12:30 – Who owns the data? Facial rights, farm telemetry, and the digital commons 15:00 – Subsurface laws: Australia’s 7-metre boundary vs. Texas property rights 16:22 – World models on the farm: Lessons from Weta Workshop and gaming engines 18:29 – Liability models: US litigation vs. NZ/Australia no-fault compensation 21:19 – The 92% safety paradox: Why we fear machines that drive better than us 23:05 – Experience-led engineering: Train Wi-Fi, elevator mirrors, and comfy cars 26:44 – Terraforming Mars: Why off-planet survival starts with farm AI 30:40 – Emerging tech: Heatwave greenhouses, mobile milking, and drone swarms 34:20 – Wrap-up & the future of physical AI Connect & Resources: Learn more about Tryfecta Capital: tryfecta.biz (https://tryfecta.biz) Connect with Titto Thomas on LinkedIn (https://www.google.com/search?q=https://www.linkedin.com/in/titto-t-b2387319/) GUESTS Melissa Clark-Reynolds ONZM, futurist, professional director and founder of Future Centre New Zealand https://www.linkedin.com/in/melissaclarkr/ (https://www.linkedin.com/in/melissaclarkr/) Nick Shelton, Director for North America, Robotyx https://www.linkedin.com/in/nickshelton/ (https://www.linkedin.com/in/nickshelton/)

3. 

   6 Sept 2026

   Ep 14: Physical AI Comes to the Farm: Melissa Clark-Reynolds on Autonomy and Ownership

   Farming is where physical AI stops being about self driving cars. Melissa Clark-Reynolds ONZM is one of New Zealand's most recognized futurists, and the room she walks into is not the one you would expect. She sits on the board of Wētā Workshop. She was Deputy Chair of Radio New Zealand, Chair of Alpine Energy, and the first independent director in the history of Beef and Lamb New Zealand. She was made an Officer of the New Zealand Order of Merit in 2015 for services to technology, after twenty five years as an entrepreneur and CEO, including building a virtual world that reached close to a million players. She trained as a futurist at the Institute for the Future in Palo Alto, the School of International Futures in the UK, and with Clayton Christensen at Harvard. She teaches governance and disruptive business models for the New Zealand Institute of Directors. She has also been early to this before. She started coding in high school and was the only person on either track of her US master's, engineering and planning on one side and epidemiology on the other, who could write code. That got her modeling pollution flows and the movement of pandemics in the 1980s. Her point about that era: the hard part was not the math, it was that there was no language yet to explain what she was doing. She thinks physical AI is in exactly that place now. She joins Titto Thomas and Nick Shelton to explain where autonomy is actually landing on farms, and the episode turns on a decision one of the world's largest equipment makers took in 2016: a corporate strategy stating they no longer wanted to sell hardware. The company famous for tractors did not want to sell tractors. What they wanted was to operate any of their equipment from anywhere, which meant digitizing farmland and betting enormously on autonomy. Her read: a new technology almost never succeeds until the operating and business model changes with it. Then the counterexample. A startup built electric autonomous tractors, and one runs a cherry orchard in New Zealand where every tree carries a code, so each cherry traces back to the tree it came from. The investment community loved the company. It could not make the business model work and the assets were sold off. Titto's line from Episode 12 was that the second mouse gets the cheese. Melissa flips it: the incumbent's balance sheet let it sweat the old assets while building the new ones, so the first mover won and the better product went broke. On why adoption stalls, her answer is not about technology. Farmers are being sold technology rather than a solution. A farmer does not want AI. A farmer cannot hire staff. Sell them a worker that runs 24 hours a day, or a machine that takes a person out of the most dangerous job in forestry, and the conversation changes. Then Nick asks the question the episode needed. What are the perils? Melissa turns it over. Farmland got mapped, and nobody asked the farmers. If drones are surveying your crops, who else can buy that data, and what happens if it shows you accidentally grew a patented seed that blew in from your neighbor. And the right to repair fight, which she compares to printer ink: if the machine reports its own fault and only the manufacturer's certified part and certified mechanic can fix it, do you still own what you bought? Melissa is back next episode for governance and world models. CHAPTERS 00:00 Introducing Melissa Clark-Reynolds 00:25 Coding in the 80s, pandemics, and the missing language 03:33 Physical AI is bigger than self driving cars 04:08 The decision to stop selling hardware 07:07 The second mouse, and why the better product went broke 09:48 Selling technology instead of selling a solution 13:01 Autonomous harvesting in Dutch greenhouses 14:25 Clarkson's Farm and how do we farm forever 15:50 Three ways to kill a weed: spray, rip, laser 19:54 Fitbits for cows and farming without fences 23:27 AI becomes a utility, like the tap 24:57 Nick Shelton on why agriculture adopts late 26:47 The perils: who mapped your farm, and who owns that data 28:45 Right to repair, and the printer ink problem 30:08 Next time: governance and world models GUESTS Melissa Clark-Reynolds ONZM, futurist, professional director and founder of Future Centre New Zealand https://www.linkedin.com/in/melissaclarkr/ (https://www.linkedin.com/in/melissaclarkr/) Nick Shelton, Director for North America, Robotyx https://www.linkedin.com/in/nickshelton/ (https://www.linkedin.com/in/nickshelton/) Host: Titto Thomas, Tryfecta Group https://tryfecta.biz (https://tryfecta.biz)

4. 

   30 Aug 2026TRANSCRIPT

   Ep 13: Beyond LLMs: Why Vision-Language-Action Models Are the Future of Physical AI

   If you are an engineer, this is the one to stay for. Prerna Dhareshwar of Voxel51 returns with Nathan Maroney to go a layer deeper than Episode 12, into what actually runs inside a physical AI system. The short version: these models predict the next action the way a language model predicts the next token, and that one architectural fact quietly dissolved a problem the industry spent years solving in hardware. Prerna works on the product side at Voxel51, where she led the platform that physical AI teams use to explore, visualize, curate and query their data. She came to it from the other end of the problem: an engineering degree from IIT Madras, a Master's from Stanford, research at India's National Aerospace Laboratories, predictive analytics at Pure Storage, and vision based anomaly detection for manufacturing at Instrumental. She has seen this from the model side, the data side, and the factory floor. Start with VLMs. They are the large language models you already know, trained on images and video alongside text, so they carry an understanding of what they are looking at. Then VLA models, Vision-Language-Action. They take every sensor input, take a language prompt describing the goal, and output action tokens continuously, each one conditioned on what the sensors are saying at that instant. Prerna walks it through with a robotic arm unloading a dishwasher. At timestamp zero it sees the dishes and decides its next move is to reach for a plate. That changes the inputs. Now the next action is to grasp. And so on. The model was never trained on your dishwasher, or Nathan's, and it does not need to be. The consequence is the most contested claim in the episode. Because the model conditions on whatever it has at each moment, the sensor streams do not need to be time synchronized. In her words, alignment of sensors is not really something people are too worried about anymore. Then trust. Titto puts the noise problem to her using his own house. A spotless kitchen is one thing. A dish sitting on the roof is another. Real environments are not controlled, and mathematically the difference is just noise. Her answer runs through post-training, the same human alignment step that makes language models sycophantic, applied instead to a human critiquing each action a robot takes. In autonomous driving that is the gap between a car that is safe and a car that behaves the way other drivers expect. Early Waymos followed the road rules exactly and got rear ended by humans who do not. Nathan names the failure mode nobody wants to discuss. Industrial pilots that succeed technically and fail commercially. Heavy industry generates enormous volumes of sensor data, but identifying the small slice worth training on takes a specialist team, and operational leaders have KPIs tied to throughput rather than to technical change. Resistance is not ignorance, it is incentives. The episode closes on Voxel51's platform, why edge cases and the long tail decide the last fraction of a percent, and why Prerna is bullish on physical AI while still calling it early. CHAPTERS 00:00 What is a VLM, and why it matters 01:36 Multimodality and where the models are heading 03:52 Sensor coverage across a mine the size of a city 04:55 Why operational KPIs block adoption 06:29 Selling a model to a board 07:55 What happens when sensors are not time synced 08:31 VLA models: next action prediction explained 09:29 The dishwasher, step by step 11:23 Why sensor fusion stopped being the problem 13:10 The world's first fully autonomous rig 13:39 Noise, uncontrolled environments, and the dish on the roof 16:51 Waymo, road rules, and getting rear ended 17:20 Industry 5.0 and human in the loop 18:12 Post-training, sycophancy, and human alignment 20:20 Manufacturing: from defect detection to assembly 22:25 Pilots that succeed technically and fail commercially 24:44 Inside Voxel51's platform 26:10 Bullish, but early 27:05 Defense, swarms, and a higher bar 30:22 Edge cases, the long tail, and the last 0.9% GUESTS Prerna Dhareshwar, Voxel51 https://www.linkedin.com/in/prernamd/ (https://www.linkedin.com/in/prernamd/) Nathan Maroney, Director, Tryfecta Group Host: Titto Thomas, Managing Partner and Co-Founder, Tryfecta Group LINKS Watch on YouTube: https://youtu.be/AFuHb-vLbyo (https://youtu.be/AFuHb-vLbyo) Voxel51: https://voxel51.com (https://voxel51.com) Tryfecta: https://tryfecta.biz (https://tryfecta.biz)

5. 

   24 Aug 2026TRANSCRIPT

   Ep 12: Beyond Vision: Why Multimodal Data and Software Power Physical AI

   Computer vision has been around for decades. So why is physical AI suddenly everywhere, with a new humanoid robotics company funded almost every week? Prerna Dhareshwar of Voxel51 and Nathan Maroney of Tryfecta Group join Titto Thomas to explain what changed, why vision alone will never be enough, and why the real differentiator is the software and data layer that almost nobody is talking about. Prerna traces the shift back to what large language models taught the field about generalization. Language modeling is next token prediction. Physical AI is next action prediction. Once it became clear the same architectures could work in both, capital and attention followed. Prerna is on the product team at Voxel51, where she led a product that helps physical AI teams explore, visualize, curate, and query their data. Before that she was a machine learning engineer building vision based anomaly detection for manufacturing at Instrumental. Nathan Maroney, Director at Tryfecta Group and the group's mining lead, brings the operator view from mining and heavy industry, where you are processing hundreds of thousands of tons and a small consistency gain in concentration is worth real money. The conversation gets practical fast. Nathan asks the question every operator is actually asking: how does a mine site get itself ready for this, and how is physical AI different from the conventional automation, machine vision, and predictive maintenance they already have? Prerna's answer is to look sideways. Auto manufacturing OEMs are already using humanoid robots to build and assemble parts, in exactly the complex three dimensional work that robotic arms could never automate away. Find what worked in an adjacent industry, then lift and generalize it. That approach also de risks the first move. On whether physical AI is just self driving cars, Prerna points out that Waymo has had more than a ten year head start collecting data, which is why that sector looks a few steps ahead of everyone else rather than being the whole story. There is an honest detour into risk. In San Francisco, Prerna keeps seeing ads for humanoid robots that will clean your house for a flat fee of $150 regardless of size. Great deal, until it leaves you with a pile of broken dishes. Titto's line for how heavy industry thinks about that: it is not the early bird that gets the worm, it is the second mouse that gets the cheese. Prerna's view is that the calculus has genuinely changed, and that she would have answered differently a year or two ago. Then the technical core. Physical AI data is hard because the sensors are not time synchronized. Each one records at its own frame rate and frequency, across long time ranges, and all of it has to be aligned, curated, labeled, and fed downstream before a model ever sees it. That tooling problem is where a lot of the real work lives. On why multimodal beats vision only, Prerna uses the human analogy. We do not perceive the world through sight alone. Two eyes give us depth. Touch tells us how much pressure a delicate object can take. Restrict a system to a single camera and you have severely limited what it can know about the world it is operating in. She takes a position on the vision only versus lidar debate, citing an edge case where a Waymo could not see a person crossing from behind a parked truck, and the lidar caught what the cameras could not. Titto brings a story from the Apache gunship program, where the sensor lens was cut from pure sapphire because the software of the 1980s could not correct for chromatic aberration. Everything had to be fixed at the physical layer to hand the software a clean image. Today a ten dollar sensor does the same job, which is exactly the hardware to software shift the field is living through. With VLMs, the sensors are fixed and the goal is fixed as a text prompt. What the system controls is the sequence of actions it takes to get there. Nathan closes on the practical blocker. Simulating a controlled separation process is achievable low hanging fruit, and most of the sensors are already installed, but a site typically has to wait two years to accumulate enough data before it can deploy. He wants systems that are self learning, self healing, and self governing from day one on a greenfield site. Prerna's answer is that it is never too early to start collecting data, and that even if you are years away from deploying anything, making sure it is clean, structured, and stored properly is the move available to you right now. She connects the self learning ambition to meta learning and few shot generalization, and to what today's models already do in a limited way through context. Prerna is coming back for a deeper technical episode on VLMs and how to deal with noise. CHAPTERS 00:00 Welcome and introducing Prerna Dhareshwar 01:06 Computer vision is not new, so what actually changed 01:22 Next token prediction to next action prediction 03:55 Nathan on mining: where the margin really sits 04:57 Lifting proven applications from adjacent industries 07:37 Is physical AI just self driving cars 09:08 The $150 humanoid and the broken dishes problem 10:33 Are we there yet, and how much should we fear it 11:56 Convincing conservative operators to move 13:46 Why software is the real differentiator 15:53 Multimodal versus vision alone 17:51 Vision only or lidar, and the Waymo edge case 18:55 Trust, and the Apache gunship sapphire lens 21:09 From hardware to software: what VLMs changed 23:10 Process simulation and the two year data wait 25:31 Start collecting data now, and the path to self learning 28:19 Next time: VLMs and filtering out noise GUESTS Prerna Dhareshwar, Product, Voxel51 https://www.linkedin.com/in/prernamd/ Nathan Maroney, Director, Tryfecta Group Host: Titto Thomas, Managing Partner and Co-Founder, Tryfecta Group LINKS Watch on YouTube: https://www.youtube.com/watch?v=o7Rx3sNpgWM Voxel51: https://voxel51.com Tryfecta: https://tryfecta.biz

6. 

   2 Aug 2026TRANSCRIPT

   Episode 11: From Mission Control to Machine Control: Inside the Physical AI Stack with Sami Sultan

   Automation does not fail at the top. It fails in the layers underneath. Sami Sultan, Vice President at Darcy Partners, ex BCG, and ex Shell wells engineer, returns for part two: a walk through the full Physical AI stack as it actually exists in heavy industry today. In this episode: The stack, layer by layer. Cameras reading rock as it comes off the shale shaker, computer vision flagging anomalies, AI agents that sense, plan, and act, and the actuation layer where a digital decision finally touches physical equipment. Human in the loop. Why the geosteer keeps their job when a few percent of production is worth millions, and why safety critical calls will stay human for a long time yet. The strange origin of mission control. How a failed military operation in 1980s Iran gave birth to joint command, then to the remote operations center, now the brain of rigs, mines, and factories everywhere. Why full stack automation ventures go bankrupt. One sensor fails and the house of cards comes down. The aerospace lesson is that automation is a management system, not a feature. Beyond the buzzword. Why Sami refuses to say digital twin, and why physics informed neural networks are winning the trust that black box AI cannot. And the next frontier. Drilling techniques crossing into mining, continuous extraction, and the bridge that could finally make Western rare earth production viable. Watch this space. About the guest: Sami Sultan is Vice President of Oil and Gas and AI at Darcy Partners, where he leads technology scouting and advisory for the world's largest energy operators and utilities. He spent nearly a decade at Shell as a wells engineer, where he helped found Shell Geodesic, an algorithmic well navigation venture that applied AI to the subsurface years before it was mainstream. He holds seven patents, is a BCG alum, and earned his MBA in sustainability from the Yale School of Management.

7. 

   20 Jul 2026TRANSCRIPT

   Ep 10: Powering the AI Boom: Distributed Energy, Microgrids, and the Future of Oil & Gas with Sami Sultan

   The AI boom has a power problem, and simply scaling up traditional energy infrastructure will not solve it. Sami Sultan, Vice President at Darcy Partners, ex Shell wells engineer, and BCG alum, joins Episode 10 to map where the energy for the machine age could actually come from: microgrids, distributed energy resources, and a pragmatic transition where existing capabilities get repurposed rather than expanded. In this episode: The power question: why energy demand compounds as AI spreads, and why the sustainable path runs through distributed generation, not just bigger grids. The transition in practice: stranded gas bridging data center demand today, old wells repurposed for energy storage tomorrow, and how capital decides what gets built. What heavy industry brings to the table: decades of experience moving liquids, running remote operations, and managing complex infrastructure, now pointed at new problems. The Physical AI stack arriving on sites today: red zone cameras, emissions drones, robot dogs, and the climb from sensing to agents to autonomy. Sami's origin story: building algorithmic well navigation at Shell in 2017, trained on synthetic wells the way Waymo trained on synthetic miles, before most boardrooms knew what a GPU was. And a first look at the three layer architecture we are building at Tryfecta: an agentic base, a physics model at the core, and command and control on top. Part one of two. Sami returns next episode for the technology deep dive.

8. 

   13 Jul 2026TRANSCRIPT

   Ep 9: Cutting Rates with Robots: Capital Flows and the Deflationary Power of Physical AI

   What if the fastest way to cut interest rates for the whole world is to teach machines to mine? Daniel Dangoor (Investments and Treasury) and Nick Shelton return for Episode 9, and this time we follow the money. Capital has been pouring into the poster children of Physical AI, drones, humanoid robots, and driverless cars, while the real prize sits underneath: machines that sense, extract, and build in the physical economy. In this episode: The deflation thesis: when Physical AI cuts the cost of mining and energy, supply rises, commodity prices fall, and the world gets easing no central bank can deliver. The iPhone economy already proved the mechanism. Why the middle of the commodity supply chain gets crushed in every cycle, and what junior miners teach us about survival. The hyperscaling question: software was the one sector that could scale 100x, and AI just ended that monopoly. Where do outsized returns come from in a physical world? Whoever has more robots wins: the case for effectively infinite capital flowing into robotics, and why debt that builds GDP is not the problem people think it is. Industry 3.0 to 6.0: from the space race that created Intel to the coming era where machines lead. The people side: why Silicon Valley is hiring problem solvers, because nobody can define an AI engineer yet. Dan closes with the best analogy of the series so far: when a person loses one sense, the others sharpen. When humanity hands its base skills to machines, watch what the remaining ones do.

9. 

   5 Jul 2026TRANSCRIPT

   Ep 8: The Compute Space Race: Geopolitics, US Hegemony, and the Physical AI Gold Rush

   How does a baby learn faster than an LLM? Not by reading more text, but by touching the world. That analogy from Daniel Dangoor (Investments and Treasury) anchors this episode's thesis: language models are capped by the finite supply of human text, and the next leap in AI depends on machines that can sense the physical world. Host Titto Thomas, Daniel Dangoor, and Nick Shelton unpack why sensors, not chatbots or humanoid robots, are the underserved gold rush of Physical AI. In this episode: Physical AI is bigger than humanoid robots and driverless cars. From rig sensors at Shell that optimized an entire fleet, to in situ soil analysis that maps rare earth deposits in a day instead of 3 months. The compute space race. Dan's macro thesis on why the US treats AI as a race it must win at any cost, and why that makes the compute investment supercycle effectively unlimited. Why sensors are the new Nvidia trade. Sensor stocks lagged every AI basket for 18 months, then rallied 80% between April and June 2026 as real industrial demand, not speculative hype, finally arrived. The ethical scaffolding. Drawing on their backgrounds in theology and philosophy, the panel asks whether governance is mature enough for the productivity and geopolitical stress ahead. Solving humanity's dirty jobs. Why machines should handle the 12 hour pipe inspections in the desert so people never have to. The takeaway: language is only the beginning. The industrial economy needs AI that can feel, and capital is now shifting to build the sensors that make that fusion possible.

10. 

    29 Jun 2026TRANSCRIPT

    Episode 7: Harnessing the Winds of Physical AI: The Change Management Blueprint for a Human-Centric Future

    While the vision of Physical AI is incredibly optimistic, the actual implementation is going to be messy. As AI native companies see productivity skyrocket, they are also facing a hidden crisis: a massive drop in workplace empathy as managers get used to bossing around AI agents 24/7 and expect the same from their human employees. In this episode of Leadership for the Physical AI Age, Nick Shelton returns to discuss the critical importance of change management as we enter the next industrial revolution. Host Titto Thomas breaks down Tryfecta's approach to building a "tag team" between human workers and AI agents, and why the ultimate human advantage will always be first-principles critical thinking. Using the tragic lesson of a 1999 Swissair crash, Titto explains why rigid, checklist-based training is dead, and how human common sense must step in to oversee the AI processes of the future. Finally, Nick and Titto revisit the ultimate question: How do we use this fire to cook our food, instead of burning down our house? In this episode, we cover: The Empathy Deficit: Why interacting with 24/7 AI agents is changing human behavior and creating friction in the workplace. The Tag-Team Model: How Tryfecta is designing systems where agents handle the heavy lifting, but call in humans for critical "last mile" process safety. The 5-Year Skill Shelf Life: Why continuous learning and education funds are the only way to survive the modern war for AI talent. The Swissair Lesson: How rigid checklists fail in crises, and why first-principles thinking is the most important skill to teach the next generation. The Good Life: Using AI to automate mundane administrative tasks to reclaim time for family, legacy building, and deep work. Learn more and connect with us: Visit our website: tryfecta.biz (https://tryfecta.biz/) Follow Titto Thomas on LinkedIn: Titto Thomas (https://www.google.com/search?q=https://www.linkedin.com/in/titto-t-b2387319/) Follow Nathan Maroney on LinkedIn: Nathan Maroney (https://www.google.com/search?q=https%3A%2F%2Fwww.linkedin.com%2Fin%2Fnathanmaroney%2F) Follow Nick Shelton on LinkedIn: Nick Shelton (https://www.google.com/search?q=https%3A%2F%2Fwww.linkedin.com%2Fin%2Fnickshelton%2F)

11. 

    22 Jun 2026TRANSCRIPT

    Ep 6: The Diamond Workforce: Hacking the Future of Jobs and AI Agents

    Titto Thomas interviews Nick Shelton, exploring how AI agents reshape the workforce into a "diamond" model, emphasizing skills for thriving alongside physical AI.

12. 

    4 Apr 2026

    Ep 5: Defending the Supply Chain: Copper, Nickel, and the New Edge

    Titto Thomas and Nathan Maroney interview each other about how essential commodities like copper and nickel fuel the Physical AI revolution and the critical need for intelligent edge architecture in mining.

13. 

    4 Apr 2026

    Ep 4: Securing the Supply Chain: Physical AI, Rare Earths, and the EV Revolution

    Tryfecta Capital co-founders Titto Thomas and Nathan Maroney explain how Physical AI secures the rare earth supply chain, de-risking investment and enabling global competition in this interview.

14. 

    4 Apr 2026

    Ep 3: Can Physical AI save mining?

    Tryfecta Capital co-founders Titto Thomas and Nathan Maroney explore how Physical AI revolutionizes mining, enabling autonomous "lights-out" operations and unlocking vast revenue potential.

15. 

    4 Apr 2026

    Ep 1: Pilot: Astro Boy, The Mongols And Accelerating The Fourth Industrial Revolution

    Host Titto Thomas interviews Tryfecta Capital co-founder Nathan Maroney about bridging the physical and software worlds to accelerate the Fourth Industrial Revolution.

16. 

    2 Apr 2026

    Ep 2: The IoT Autopsy: Why 10 Years of Sensors Didn't Change the World

    Host Titto interviews Jason Quarberg, Director of AI and Technology at Tryfecta Capital, dissecting why IoT failed and how Physical AI now enables autonomous action on industrial data.

Inside a recent episode

Ep 13: Beyond LLMs: Why Vision-Language-Action Models Are the Future of Physical AI

Published 30 Aug 2026 · Transcript excerpt

[…] Definitely. I think a large part of convincing people that AI is valuable is also just education around what is possible with physical AI, right? I think unless it's more widely understood and unless people see that there's value in using physical AI in various different industries, it's unlikely you would want to be trying out a completely unknown technology that might be risky otherwise because you know that something already works. Why would you want to now go try something that's not proven out and not known to have value and take that risk? I think education goes a long way also. So some amount of understanding how these systems work, understanding where they've had success, where they've had value, where they've had impact helps you decide whether you want to try adopting that technology in a similar fashion in your own job and your own use case. […]

Contact Leadership for the Physical AI Age

Guest appearances

Books guests

Based on episode analysis; this does not confirm that the show is currently accepting guests.

Questions about Leadership for the Physical AI Age