Episodes

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Damian Borth, a professor of artificial intelligence and machine learning at the University of St. Gallen and CEO of Weight Space Labs, about a different way to build AI models. Training a neural network usually takes data, computing power, and time. But what if neural networks could also learn from the millions of models that have already been trained? Damian explains weight-space learning in simple terms. A neural network stores what it learns in its weights. His research asks whether another model can read, edit, or use those weights to generate a new one. The goal is to create smaller, task-specific models without repeating every step of training. Damian describes a remote-sensing study in which his team generated models for tasks such as classifying satellite images and detecting wildfires or floods. He says comparable models in that study needed 20 times less compute than models trained from scratch.…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Jake Steinerman, a spatial computing practitioner working in developer relations at Cesium, about how computers connect digital objects and models to real places. The question is more practical than it sounds: how does a phone know where to place an augmented reality arrow when someone walks down an unfamiliar street? Jake explains that spatial computing is not one device or one technology. Augmented reality can place content over what a person sees; virtual reality can put them inside a digital environment. A phone can deliver useful AR without a headset. To keep an object in the right place, an app may recognize an image, the shape of a machine, the surfaces of a room, or the buildings seen through a camera. Outdoor navigation can combine location, direction, and visual matching rather than relying on GPS alone. Vit and Jake then move from immersive experiences to the data behind them. Geospatial tools p…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Marco Benitez, co-founder and CEO of ROOK and a biomedical engineer, about why a wearable can track months of health data while a doctor may still see only a snapshot at the next visit. Marco explains the gap between collecting readings and using them in care. Sleep, activity, blood pressure, and other information can sit in separate device apps, lab portals, medical systems, and electronic health records. ROOK builds infrastructure for digital-health platforms to connect those sources; it is not a direct doctor-facing app. Even after a connection works, teams must normalize formats, handle duplicate data, and decide which measurements are reliable enough for a given use. Vit asks what a clinician actually needs from that longer view. Marco argues that sending every raw reading would add work to an already busy practice. Relevant trends and alerts may fit better, particularly in chronic-disease management,…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Raffi Kryszek, a Chief Product and AI Officer at Proto Hologram, about what conversational AI holograms actually do and when a digital person can be more useful than a chatbot, app, or ordinary screen. Raffi explains that the word “hologram” can describe several different experiences. Proto began with live remote presence: making a person appear in a shared physical space. The same display can also show prerecorded content, an interactive application, or an AI persona. Vit and Raffi clarify why Proto built an operating system alongside its hardware. The conversation follows a question from someone's voice through speech recognition, a language model and connected data, generated speech, and an avatar. That chain helps explain both the appeal and the limits of AI with a face. A natural-looking response is not necessarily a reliable one; the assistant still needs accurate information, sensible guardrails, an…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Jeff Keyes, Field CTO at Allstacks and a product and go-to-market leader, about why AI can help product teams build the wrong thing faster when companies measure output instead of customer value. Vit and Jeff discuss the pressure many software organizations now feel to prove ROI from AI tools. Jeff explains why the easiest metrics, more lines of code, more pull requests, more tickets, more user stories, and more shipped features, can be misleading. A team can produce more work and still fail to improve adoption, usage, retention, revenue, or the customer experience. The conversation explores what AI slop looks like inside real product teams. Jeff argues that AI-generated requirements can become the same weak user stories with four times more text, and that polished documents do not automatically mean clearer thinking. He connects this to product management discipline: understanding the customer problem, de…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Tammuz Dubnov, co-founder and CTO of AutonomyAI, about what happens when AI coding agents move from impressive demos into real production products with legacy code, dependencies, architecture, testing, review queues, and product judgment. Vit and Tammuz discuss why building a small app with AI is very different from changing an existing codebase. Tammuz explains how AI-generated pull requests can look finished to a product manager but still be too large, unmaintainable, or hard for engineers to merge. The conversation separates demo magic from production reality: a feature is not truly done just because it appears to work. They explore why the harness around an AI agent matters as much as the model itself. Tammuz describes how a product-focused agent needs to translate intent, find the right files, avoid irrelevant code changes, run checks, test the product visually, and verify that the result matches the …

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Yue Zhao, product leadership coach and creator of The Uncommon Executive, about why strong product managers can struggle when they move into leadership and what has to change before, during, and after that transition. Vit and Yue discuss how a PM can tell whether they are ready for product leadership. Yue argues that readiness is not only about ambition. A stronger signal is whether people already come to that PM for judgment and help beyond their formal area of responsibility. When others pull someone into broader problems, they may already be acting at the next level. The conversation explores why the habits that make someone a reliable PM can become liabilities as a leader. Product managers are often rewarded for solving problems directly, tracking details, and making sure work gets done. Product leaders, by contrast, need to create systems, coach people, build relationships, and make only the decisions…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Rob Gramlich, founder and president of Grid Strategies, about AI data centers, electricity demand, the electric grid, transmission bottlenecks, and why power may become one of the biggest constraints on AI growth. Vit and Rob discuss Grid Strategies’ forecast that U.S. power demand could grow by around 166 gigawatts by 2030, a scale Rob compares to about 15 New York Cities worth of power. They break down how the grid works across generation, high-voltage transmission, and local distribution, and why the U.S. effectively operates through three major grid systems: Eastern, Western, and Texas. The conversation explores why AI data center developers often want power faster than utilities can safely provide it, why transmission lines are difficult to build, and why transformers, turbines, breakers, copper, rare earth minerals, and other equipment have become scarce and expensive. Rob also explains why utilities…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Louis Rosenberg, CEO and Chief Scientist of Unanimous AI, about collective intelligence, swarm intelligence, AI-mediated group decision-making, and why AI’s best use may be connecting people rather than replacing them. Louis explains why polls, surveys, and votes can aggregate opinions but often fail to help a group deliberate, challenge assumptions, or discover the strongest answer. Vit and Louis discuss how nature solves group intelligence through fish schools, bird flocks, and bee swarms, and why humans are good at thoughtful small-group conversation but struggle when a conversation grows to hundreds or thousands of people. Louis explains how AI agents can observe many small conversations, carry insights between groups, and weave large networks of people into one real-time deliberation. The conversation explores evidence from IQ-test research, radiology diagnosis, forecasting, and problem-solving, inclu…

In this episode of the An Hour of Innovation podcast, Vit Lyoshin speaks with Jolie Dobre, author of Sustaining Human Intelligence: Keeping People Smart, Connected and Innovative in an AI-Driven Workforce, about how AI can speed up work while quietly weakening the human skills organizations depend on. Jolie explains cognitive offloading, skill decay, and why generative AI differs from narrow tools like GPS and calculators: it can take over summarizing, reasoning, ideation, research, and judgment across many domains. Vit and Jolie discuss how individuals can tell whether AI helps them think more deeply or simply does the thinking for them. Jolie shares a practical delegation test: before handing work to AI, ask whether you understand the task, the problem behind it, the goal, and when you last did the work yourself. The conversation also explores why junior employees are especially vulnerable if AI automates away foundational tasks before they build pattern recognition and deliberat…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Jonathan Schaeffer, founder and CEO of Synsira and creator of Kind, about why today’s AI tools can be useful but still unsafe to trust blindly. Jonathan explains why large language models such as ChatGPT, Gemini, and Claude should not be treated as intelligent authorities, why the word “hallucination” makes fabricated answers sound too harmless, and why a polished AI response can still contain errors users may never catch. Jonathan shares how ChatGPT once generated a biography about him with dozens of mistakes, then appeared to improve later through what he calls band-aids rather than a deeper solution to the trust problem. He explains why training on broad internet data can produce strange or unsupported answers, and why curated datasets can reduce fabrications by grounding responses in information the user controls. The conversation also covers source citations, verification, and why trustworthy AI should b…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Rebecca Rosenberg, founder of ReBokeh, about why low-vision accessibility is not niche and what it teaches product teams about adoption and business value. Rebecca explains how her low vision from albinism led her to build a visual-first app that uses a smartphone camera, augmented reality, contrast, brightness, color filters, and AI descriptions to help people customize what they see in real time. Rebecca argues that products often fail because teams build for the wrong user, market to the wrong user, or skip early conversations that reveal what problem actually matters. Her work with ReBokeh shows why accessibility has to be designed around real users rather than assumptions. Low vision is not simply “less blind”; it is a distinct experience where people may still rely heavily on sight but need different tools than audio-only or screen-reader-first solutions. Vit and Rebecca discuss the overlooked size o…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Sid Bharath, founder of Refound, about why many AI projects fail to produce measurable ROI and what companies should do before giving work to agents. Sid argues that the problem is often not the model alone, but the way companies approach automation: large AI makeovers, unclear outcomes, incomplete data, missing edge cases, and workflows that were never documented well enough to automate. Sid explains what becoming AI-native actually means in everyday business operations. It does not mean removing humans or handing every decision to AI. It means identifying the repetitive operational layer of work, giving agents enough context and access to execute defined tasks, and keeping people focused on judgment, strategy, creativity, relationships, and review. Vit and Sid discuss how this changes roles inside sales, operations, engineering, customer support, accounting, and startup teams. The conversation also cover…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Ravi Sawhney, founder and CEO of RKS Design and creator of the Psycho-Aesthetics methodology, about why products fail even when the technology works. Ravi explains that product adoption is not only about function, usability, or technical performance; it is also about trust, stigma, identity, emotional resonance, and how a product makes people feel about themselves. Ravi shares a powerful story from his work on the MiniMed diabetic infusion pump, a product that improved quality and length of life but struggled because people felt exposed by its medical appearance in public. That experience helped shape Ravi’s view that successful design must understand the person using the product, the social meaning of adoption, and the emotional signal a product sends. Vit and Ravi also discuss Psycho-Aesthetics, persona resonance, the Hero’s Journey, self-affirmation, and how teams can use deeper human insight to avoid b…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Sergii Gerasymovych, founder, CEO, and an AI data center infrastructure operator at EZ Blockchain, about why the next major AI bottleneck may not be models or GPUs, but the physical infrastructure required to power them. Sergii explains how EZ Blockchain is pivoting from Bitcoin mining infrastructure into AI data centers, and why the same lessons from crypto mining now apply to the AI compute race: everyone wants hardware online quickly, but land, power, utilities, transformers, cooling, and construction move much slower than software. The conversation breaks down what an AI data center actually is, why power and land sit underneath the entire compute stack, and why chip supply may be easier to scale than electricity delivery. Sergii argues that companies can manufacture more GPUs, TPUs, and specialized chips, but they cannot print power, accelerate utility studies overnight, or build substations at startup s…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Dr. Shohini Ghose, quantum physicist, author, and professor at Wilfrid Laurier University, about what quantum computing actually is and why the most urgent risk may not wait for a perfect large-scale quantum computer. Shohini explains why quantum computing is not simply a faster version of today’s computers. It starts from qubits instead of classical bits, which means concepts like superposition, entanglement, and probability become part of the computing model itself. Vit and Shohini discuss why governments and companies are investing heavily in quantum technologies, why the science is solid but scalability is still difficult, and why most people probably will not own a personal quantum computer anytime soon. Instead, quantum access may look more like cloud infrastructure or specialized services used when a problem has a real computing, sensing, security, or communications bottleneck. They also explore whe…

In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Ethan Barajas, co-founder and CEO of Icarus Robotics, about why the space economy may be limited less by rockets and more by labor in orbit. Ethan explains why astronauts spend valuable time on cargo movement, maintenance, experiment setup, and routine station operations, and why that work becomes a major bottleneck as commercial space stations, orbital manufacturing, and future space infrastructure grow. Vit and Ethan discuss how Icarus Robotics is building robotic workers for space stations, how teleoperation can become the data layer for autonomy, and why AI robotics in microgravity is very different from robotics on Earth. They also explore what it takes to build a company for a market that is still forming, how founders validate demand before building, and why robot labor could help astronauts focus on science instead of housekeeping. The conversation covers space robotics, commercial space stations, …