#261 Jonathan Frankle: How Databricks is Disrupting AI Model Training
This episode is sponsored by Oracle. OCI is the next-generation cloud designed for every workload – where you can run any application, including any AI projects, faster and more securely for less. On average, OCI costs 50% less for compute, 70% less for storage, and 80% less for networking. Join Modal, Skydance Animation, and today’s innovative AI tech companies who upgraded to OCI…and saved. Try OCI for free at http://oracle.com/eyeonai What if you could fine-tune an AI model without any labeled data—and still outperform traditional training methods? In this episode of Eye on AI, we sit down with Jonathan Frankle, Chief Scientist at Databricks and co-founder of MosaicML, to explore TAO (Test-time Adaptive Optimization)—Databricks’ breakthrough tuning method that’s transforming how enterprises build and scale large language models (LLMs). Jonathan explains how TAO uses reinforcement learning and synthetic data to train models without the need for expensive, time-consuming annotation. We dive into how TAO compares to supervised fine-tuning, why Databricks built their own reward model (DBRM), and how this system allows for continual improvement, lower inference costs, and faster enterprise AI deployment. Whether you're an AI researcher, enterprise leader, or someone curious about the future of model customization, this episode will change how you think about training and deploying AI. Explore the latest breakthroughs in data and AI from Databricks: https://www.databricks.com/events/dataaisummit-2025-announcements Stay Updated: Craig Smith on X: https://x.com/craigss Eye on A.I. on X: https://x.com/EyeOn_AI
--------
52:47
#260 Ash Anwar: How Molecular You Uses AI for Early Disease Detection
AGNTCY - Unlock agents at scale with an open Internet of Agents. Visit https://agntcy.org/ and add your support. What if your blood could reveal the future of your health - years before symptoms ever appear? In this episode of Eye on AI, Craig Smith sits down with Ash Anwar, co-founder of Molecular You, to explore how AI and advanced biomarker analysis are transforming how we detect and prevent disease. From a real-life case where stage 1 pancreatic cancer was caught early, to the science behind tracking 250+ dynamic biomarkers, Ash breaks down how Molecular You is shifting healthcare from reactive treatment to proactive longevity. They dive into the limitations of genetic testing, the power of machine learning models trained on clinical data, and how personalized action plans are helping individuals take control of their health in real time. If you're curious about the intersection of AI, diagnostics, and the future of preventive medicine—this is a conversation you won't want to miss. Stay Updated: Craig Smith on X: https://x.com/craigss Eye on A.I. on X: https://x.com/EyeOn_AI (00:00) How AI Can Predict Disease Before It Happens (02:02) Meet Ash Anwar: From Scientist to Health Tech Leader (06:31) What Is Molecular You and How It Works (09:43) Why Biomarker Tracking Over Time Is a Game-Changer (15:46) How Molecular You Detected Stage 1 Pancreatic Cancer (23:15) Biomarkers vs Genetics: What Really Matters (28:11) The AI Models Behind Early Disease Detection (32:37) How the Product Works for Clinics and Consumers (35:05) Who They Compete With and What Makes Them Different (37:34) Research vs Product: The Cancer Risk Challenge (43:43) How Often Should You Get Tested?
--------
49:13
#259 Anjney Midha: a16z’s Strategy to Turn AI Startups into Unicorns
AGNTCY - Unlock agents at scale with an open Internet of Agents. Visit https://agntcy.org/ and add your support. What does it really take to turn cutting-edge AI research into a successful foundation model company? In this episode of Eye on AI, we sit down with Anjney Midha, General Partner at a16z, to unpack how he helps scientists and researchers transform their breakthroughs into scalable, real-world AI businesses. From his early backing of Anthropic to launching Mistral and Black Forest Labs, Anjney shares a behind-the-scenes look at how AI infrastructure companies are born. We dive into the critical challenges of model reliability, evaluation beyond academic benchmarks, and the rise of hybrid architectures combining transformers with diffusion and LSTMs. If you're building in AI or investing in it, this is the roadmap to what's next. Stay Updated: Craig Smith on X:https://x.com/craigss Eye on A.I. on X: https://x.com/EyeOn_AI (00:00) Turning AI Research into Real Companies (02:08) Anjney’s Journey into Venture Capital (05:44) The Birth of Anthropic (08:26) Backing Mistral and Stable Diffusion (13:16) Are Transformers Really Enough? (18:36) Why AI Evaluation Is Broken (22:10) Making AI Models More Interpretable (28:38) The Real Potential of AI Agents (32:43) How a16z Spots AI Breakthroughs (37:45) Investing Like It’s the 1970s (43:31) What AI Voice Tech Needs Right Now (46:32) Models vs Products (51:17) What’s Holding Back AI Agents (55:41) Anjney Startup Investing Strategy
--------
56:38
#258 Brian Peterson: How Dialpad is Building the Future of AI-Powered Communication
What happens when you combine 8 billion minutes of voice data with a full-stack AI engine? Yes, that’s what Dialpad is doing. In this episode, Brian Peterson, CTO and Co-Founder, breaks down how they’ve built an AI-powered communications platform from the ground up. From real-time sales coaching and AI-driven support agents to predictive analytics that can spot churn before it happens, Brian shares why owning the full stack — infrastructure, LLMs, and data is the only way to deliver truly intelligent customer experience. If you’re curious about the future of AI in business communication, this is the episode to watch. Stay Updated: Craig Smith on X:https://x.com/craigss Eye on A.I. on X: https://x.com/EyeOn_AI (00:00) Brian’s Founding Story (03:56) What Dialpad Actually Does Today (05:17) Is Voice the Most Valuable Untapped Data Source? (07:41) Inside DialpadGPT (10:10) AI Solutions for Sales, Support & Collaboration (12:24) Owning the Entire Customer Journey with Unified Comms (14:11) How Dialpad Stays Ahead in the AI Race (17:50) Real-Time AI Coaching & Playbooks (22:32) Why Most Enterprises are Behind in AI Adoption (25:28) Action-Oriented AI Agents (32:40) What’s Next for AI in Customer Communication
--------
36:44
#257 Ankur Banerjee: How cheqd.io Is Building Trust for AI Agents with Decentralized Identity
AGNTCY - Unlock agents at scale with an open Internet of Agents. Visit https://agntcy.org/ and add your support. What if AI agents needed digital passports to act on your behalf? In this episode, Ankur Banerjee, Co-Founder and CTO of cheqd.io, reveals how decentralized identity is becoming the foundation of the AI agent economy. From booking Taylor Swift tickets with an agent to proving you're a real person online, we explore why identity and trust are the hidden infrastructure shaping the future of AI. Ankur explains how cheqd is building privacy-first tools that let AI agents verify who they are, what they can do, and who they're working for, all without handing over your data to big tech. We dig into the rise of digital credentials, the limits of biometrics, and how protocols like MCP are making the internet safe for autonomous agents. If you've ever wondered how AI will operate on your behalf in the real world, this conversation offers a glimpse into what's coming next. Stay Updated: Craig Smith on X: https://x.com/craigss Eye on A.I. on X: https://x.com/EyeOn_AI (00:00) Why AI Needs Identity (02:16) Ankur’s Path to cheqd (05:39) Delegating Tasks to AI Agents (11:26) Identity Lessons from Governments (17:25) Trusting AI with Digital Credentials (23:42) The Problem with Biometrics (30:49) Web Standards for Agent Identity (36:46) MCP and Agent Interoperability (48:47) Bridging Web2 and Web3 Identity (55:18) Why Companies Should Care About Decentralized ID
Eye on A.I. is a biweekly podcast, hosted by longtime New York Times correspondent Craig S. Smith. In each episode, Craig will talk to people making a difference in artificial intelligence. The podcast aims to put incremental advances into a broader context and consider the global implications of the developing technology. AI is about to change your world, so pay attention.
Österreich