The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Sam Charrington
Machine learning and artificial intelligence are dramatically changing the way businesses operate and people live. The TWIML AI Podcast brings the top minds and ideas from the world of ML and AI to a broad and influential community of ML/AI researchers, data scientists, engineers and tech-savvy business and IT leaders. Hosted by Sam Charrington, a sought after industry analyst, speaker, commentator and thought leader. Technologies covered include machine learning, artificial intelligence, deep learning, natural language processing, neural networks, analytics, computer science, data science and more.
Episodes
Mentioned books
81 snips
Oct 28, 2026 • 54min
The Brutal Reality of Building Agentic Products Today with Daniel Jeffries - #782
Daniel Jeffries, technology writer, entrepreneur, and creator of Nautilo, explores the realities of building useful AI agents. They discuss personalized and collaborative agents, persistent memory, deterministic tools, privacy-focused model routing, multi-agent systems, and the road to local AI. The conversation also tackles monetization, differentiation, distribution, and why co-creative workflows may outperform “prompt and pray” approaches.
97 snips
Oct 21, 2026 • 52min
Airbnb Is Building a ‘Company Brain’ for AI Agents with Ahmad Al-Dahle - #781
Ahmad Al-Dahle, Airbnb’s CTO and a former Apple iPhone builder and Meta Llama leader, discusses Project Everest, a company brain for AI agents that captures institutional knowledge. They explore stale context, agent evaluations, measuring productivity beyond tokens, support automation, multimodal search, and the thorny challenge of resolving disputes in a two-sided marketplace.
62 snips
Oct 15, 2026 • 53min
Your Ranking Model Is Optimizing the Wrong Thing with Zoe Weil - #780
Zoe Weil, co-founder and CEO of Sequen, builds real-time ranking systems for consumer businesses. She discusses why clicks and search relevance can miss purchase intent, how hovers, scroll speed, and dwell time shape live recommendations, and why retail needs different methods than social platforms. The conversation also explores anonymous personalization, Rust-powered millisecond ranking, specialized model “orchestras,” and improving ad and organic revenue together.
312 snips
Oct 6, 2026 • 1h 30min
Why Jev Is Changing How We Build With AI with Diogo Almeida - #779
Diogo Almeida, TypeSafe co-founder and CEO and former OpenAI researcher, discusses Jev, a model built for reliable machine-native intelligence. They explore why language models struggle with routine automation, how calibrated decisions and RLCD could make AI probabilities more trustworthy, and why modular software engineering may outperform one-model agents. The conversation also examines Jev’s potential as an intelligence API for agents, tool use, and AI-powered software.
92 snips
Sep 29, 2026 • 1h 8min
From Math Olympiads to Navier-Stokes: How Fast Is AI Progressing? with Greg Burnham - #778
Greg Burnham, an AI capabilities researcher leading research at Epoch AI, explores how machines advanced from grade-school math to frontier problems like Navier-Stokes. They discuss measuring progress beyond saturated benchmarks, whether AI can generate genuinely new mathematical ideas, and why open-ended research and learning from experience remain difficult. The conversation also examines compute scaling, self-improvement, board-game tests, and potential real-world capability thresholds.
35 snips
Sep 17, 2026 • 1h 5min
From Voice Agents to AI Avatars with Alexander Smola - #777
Alexander Smola, co-founder and CEO of Boson AI and a Carnegie Mellon professor, explores the leap from voice agents to audiovisual AI avatars. He discusses real-time audio, latency, tokenization, scalable training data, and architectures that balance intelligence with responsiveness. The conversation also examines emotional intelligence, natural interruptions, personalization, memory, and how AI systems might learn social skills through interaction.
170 snips
Sep 9, 2026 • 59min
Do AI Tokenomics Matter More Than Model Benchmarks? with Chris Potts - #776
Chris Potts, a Stanford linguist and AI researcher, explores tokenomics, tokenflation, and why benchmarks miss the bigger picture of AI progress. He discusses inference-time scaling, efficient transformer architectures, DSPy’s modular approach, interpretability, and how AI fluency shapes results. The conversation also looks toward recursive models, byte-level systems, open weights, and new paths beyond gradient learning.
86 snips
Sep 1, 2026 • 1h 6min
World Models and the Future of Spatial AI with Justin Johnson - #775
Justin Johnson, computer scientist, University of Michigan associate professor, and World Labs co-founder, explores world models and spatial AI. He unpacks competing definitions, POMDPs, reinforcement learning, and the tension between explicit 3D representations and generative models. He also discusses Marble’s navigable 3D worlds, training with images and video, evaluating consistency, and the path toward AI that can simulate, plan, and act in virtual and physical environments.
72 snips
Aug 26, 2026 • 58min
Why the Next AI Breakthrough May Come from Physics with Max Welling - #774
Max Welling, professor of machine learning and CuspAI co-founder, pioneers equivariant networks and AI-driven materials discovery. He discusses generative AI for designing semiconductors, batteries, and carbon capture materials. He explores physics-inspired ideas for AI, including thermodynamics, waves as a computational primitive, and symmetry breaking to reshape future architectures.
47 snips
Aug 12, 2026 • 57min
Why Image Generation Needs More Than Bigger Models with Fatih Porikli - #773
Fatih Porikli, VP of Technology at Qualcomm and computer vision researcher, discusses shortcomings of current image generation and why realism is not correctness. He covers controllability, decomposing planning from rendering, methods for identity-diverse outputs, on-device 16MP generation, and artifact-free image editing in multiple short, focused conversations.
