The TWIML AI Podcast with Sam Charrington

The Brutal Reality of Building Agentic Products Today

1.9K views · 5d ago · 53:32

AI agents are becoming accessible to everyone, raising the bar for developers building the next generation of agent products. When capable, general-purpose agents are available out of the box, differentiation increasingly comes down to the architectural decisions and specialized experiences that make an agent uniquely valuable.

Topics: The agency problem · Nautilo · Multi-agent systems · Context window limitations · Tool design philosophy: "Magical Hammer Design” · Venice

How Airbnb Prevents Stale Knowledge From Poisoning Agent Context

2.6K views · 1w ago · 51:18

AI agents need the institutional knowledge experienced employees rely on: why a system works a certain way, which policies apply, and what never made it into the documentation. Capturing that knowledge is hard. Keeping it useful as people and agents add more may be harder.

Optimizing AI Recommendations for Revenue with Sequen Founder Zoe Weil

1.8K views · 2w ago · 52:15

Guest: Sequen Founder Zoe Weil

In this episode, Zoe Weil, co-founder and CEO of Sequen, joins us to discuss how ranking systems can use live behavioral signals to optimize for purchases rather than clicks or search relevance alone. Zoe explains why approaches developed at data-rich social platforms don’t necessarily translate to retailers and how Sequen uses scroll speed, hover duration, and other in-session interactions to infer shopper intent without relying on user identity.

Topics: Sequin’s Ranking Intelligence Layer · Difference with Search and Relevance · Session Live Signaling · Surveillance Capitalism · Building a Low-Latency Stack in Rust · Model Architecture

Asking JEV's Creator the Hard Questions About This Hot New AI Model

10K views · 3w ago · 1:29:40

In this episode, Diogo Almeida, co-founder and CEO of TypeSafe, joins us to discuss Jev, TypeSafe’s recently released model for bringing fast, reliable intelligence directly into software. We explore the idea of “machine-native intelligence” and why Diogo believes models optimized for generating text are poorly suited to many of the decisions required for real-world automation.

Topics: Hot Takes · The Bitterest Lesson · Diogo’s Career Background · Sci-Fi Software Demos vs. Easy Unsolved Problems · Jev · How Jev Differs from Classifiers and Small Models

AI Progress Explained: From Math Olympiads to Navier-Stokes

2.7K views · 1mo ago · 1:07:19

AI systems have gone from struggling with grade-school math to helping solve research problems that have resisted mathematicians for decades, including Navier-Stokes. In this episode, Greg Burnham, who leads AI capabilities research at Epoch AI, joins us to examine what that progress says about where AI is going.

Topics: AI Capabilities Research · Is AI Progress Slowing Down? · Correlation and Linear Trend of Benchmarks · Drivers of AI Progress · Evolution of AI Math Capabilities · How AI Solves Hard Math Problems

From Voice Agents to AI Avatars

2.5K views · 1mo ago · 1:04:29

Voice AI has gotten remarkably good, but natural conversation remains a high bar. Small delays, awkward interruptions, or the wrong tone can quickly break the illusion, and adding vision and visual presence only raises the stakes.

Topics: Challenges of Voice AI in Real-World Environments · Why Voice Systems Are a Whole-System Challenge · Trade-Off Between Size, Compute, and Affordability · Boson AI’s Path to Audio · Training Dataset · Dealing with Noisy Data

Tokenomics: What is Your AI Dollar Really Buying You?

2.4K views · 1mo ago · 59:00

As reasoning models consume more tokens and AI systems become more expensive to run, understanding what those tokens actually buy is becoming increasingly important. In this episode, Stanford professor and Big Spin co-founder Chris Potts joins us to discuss AI tokenomics and his research into “tokenflation”—the possibility that token usage is growing faster than the measurable value those tokens produce.

Topics: Linguistics in the Age of Language Models · Scale Limitations in NLP Research · Challenging the Bitter Lesson Mindset · Relationship Between Data, Mechanistic Interpretability, and Efficiency · DSPy · Prompt Optimization and Model Variability

Are World Models What Comes After Language Models?

4.9K views · 2mo ago · 1:05:31

In this episode, Justin Johnson, co-founder of World Labs, joins us to discuss world models and the emerging field of spatial AI. We explore why many researchers see capabilities beyond language as an important frontier for AI, and what it means to build models that can understand, generate, and simulate the environments around them.

Topics: Defining World Models · World Models as Theory Builders · POMDPs and Agent–World Interaction · Training Agents with Behavior Cloning · Ground-Truth State and Learned State · Explicit 3D vs. Implicit 3D

Why the Next AI Breakthrough May Come from Physics

4.4K views · 2mo ago · 57:36

The conventional wisdom in AI is that the next breakthrough will come from more compute, more data, and larger models. But what if the next leap comes from somewhere else? In this episode, Max Welling—co-founder and CTO of CuspAI and professor at the University of Amsterdam—argues that physics may provide some of the ideas behind the next generation of AI systems.

Topics: From Equivariant Networks to AI for Science · Founding CuspAI · Using AI to Discover New Materials · Designing Materials for Carbon Capture · Partnerships and the CuspAI Business Model · The End-to-End Materials Discovery Process

AI Image Generation Is Not As Good As You Think It Is

2.3K views · 2mo ago · 56:18

Text-to-image models have become remarkably good at producing realistic images. But realism isn’t the same as correctness. Ask for several distinct people, a specific composition, or a high-resolution image generated locally, and today’s models still struggle in surprising ways.

Topics: What’s Still Unsolved in Image Generation · Disco: Improving Identity Diversity · Better Training Objectives and Reinforcement Learning · Are We Asking One Model to Do Too Much? · Agentic Image Generation Pipelines · Evaluating Diversity with the DiverseHumans Benchmark

Meet The Researcher Unlocking Knowledge Trapped In LLMs

3K views · 3mo ago · 46:31

For more than a decade, AI has advanced by training ever-larger models on ever-larger datasets. But as high-quality training data becomes harder to find and pretraining grows increasingly expensive, researchers are looking for new ways to keep foundation models improving.

Topics: Research focus · Weight space learning · Symmetries and transformations in weight space · Training on the Hugging Face Model Zoo · Filtering models · Tokenization and normalization

AI Is Unlocking the Sense of Smell

3.2K views · 3mo ago · 59:26

In this episode, Alex Wiltschko, founder and CEO of Osmo, joins the show to discuss his goal of giving computers a sense of smell and what it takes to build olfactory intelligence. We explore the science behind smell, from the hundreds of olfactory receptors in the human nose to the challenge of mapping the relationship between molecular structure and odor, ensuring safety regulations are met, and building foundation models for smell.

Topics: Background · Digitizing smell · How powerful is the human nose? · Structure-odor relation problem · Nodes and edges of graph neural net · Principal odor map

AI Factories Explained: What’s Actually New?

4.4K views · 4mo ago · 15:25

Guest: Thierry Pienaar

AI factory is one of those terms that seemed to appear overnight. Vendors are talking about it. Enterprises are investing in it. Governments are funding it. But what is an AI factory, really? In this episode of the TWIML Briefing Room, recorded live at HPE Discover 2026, I sit down with Thierry Pienaar, CTO of AI & HPC at HPE, and Kaushik Shirhatti, VP of AI Factory at NVIDIA, to unpack what’s actually changing in enterprise AI infrastructure.

Topics: The 5 Layers of the AI Factory Stack · Private Cloud vs. AI Factory: Scaling for the Enterprise · The Shift from Training to Inference & Cost Optimization · Enterprise Challenges: Data, ROI, and Prioritization · Optimizing vs. Reimagining Workflows · On-Prem vs. Cloud: Why AI is breaking down IT silos

Can This Tiny Board Replace My Al Agent Server?

2.2K views · 4mo ago · 8:25

Can this tiny computer a replace a cloud VPS for AI agents? In this video, I put that idea to the test using the Rubik Pi 3, a Qualcomm Dragonwing-powered single-board computer, and OpenClaw, an open-source AI agent framework.

If Guardrails Can't Make Your Agents Behave, What Will?

3.9K views · 4mo ago · 55:49

Guest: Dev Rishi

In this episode, Sam talks with Dev Rishi, GM of AI at Rubrik, about what happens when agents move beyond answering questions and start taking action across tools, systems, and business processes. We explore why the enterprise playbook of static guardrails plus human approval starts to break down in the agent era.

Topics: Barriers to enterprise AI adoption · Rubrik and defining agents · Limitations of human-in-the-loop and legacy security · Zero trust in agents · Three pillars of agent security · SAGE