The TWIML AI Podcast
Formerly known as This Week in Machine Learning & AI, this podcast brings top minds and ideas from the world of machine learning and artificial intelligence to a broad audience. Sam Charrington interviews ML/AI researchers, data scientists, and industry leaders about the latest innovations and trends, making AI insights accessible to practitioners and business leaders alike.
Is RAG Dead? Lessons From Building AI for Tax Law with Alex Bowcut - #769
In this episode Sam Charrington talks with Alex Bocut, Head of Engineering at Sphere, about the role of Retrieval‑Augmented Generation (RAG) in AI‑driven tax compliance. Bocut explains why, for highly regulated domains like sales‑tax automation, pure large‑context models aren’t enough—accurate citations and legal traceability are essential, so Sphere relies on a hybrid RAG system called TRAM that retrieves relevant statutes, translates them, and feeds them to a specialized model. He describes how TRAM accelerates tax experts’ review workflow by up to two orders of magnitude while maintaining accuracy, and how the pipeline handles diverse source formats (HTML, PDFs, images, spreadsheets) and multilingual legislation. The discussion highlights the practical limits of “RAG is dead” arguments and showcases a real‑world AI application that blends retrieval, translation, and expert validation.
Relational Foundation Models for Enterprise Data with Jure Leskovec - #768
In this episode, Sam Charrington talks with Jure Leskovec, co‑founder and chief scientist of Kumo and a Stanford professor, about his work on relational foundation models that can reason directly over structured enterprise data without any task‑specific training. Leskovec explains...
The Race to Production-Grade Diffusion LLMs with Stefano Ermon - #764
In this episode, host Sam Charrington talks with Stefano Ermon, a Stanford associate professor and CEO of Inception, about the evolution of diffusion models from image generation to text and code. Ermon explains how diffusion models, which iteratively denoise from...
Why Vision Language Models Ignore What They See with Munawar Hayat - #758
In this episode, Qualcomm AI Research scientist Munawar Hayat explains why Vision‑Language Models often ignore visual input, leading to object hallucination, and how his team’s attention‑guided alignment technique improves visual grounding. He also introduces Generalized Contrastive Learning for efficient multi‑modal...
Proactive Agents for the Web with Devi Parikh - #756
In this episode, Devi Parikh, co‑founder and co‑CEO of Yutori, explains how proactive web agents can transform browser interactions by using visually‑grounded models that operate on screenshots instead of the fragile DOM, resulting in more robust handling of complex interfaces....
AI Orchestration for Smart Cities and the Enterprise with Robin Braun and Luke Norris - #755
In episode #755, Robin Braun of HPE and Luke Norris of Kamiwaza explore how AI orchestration can automate complex workflows for smart cities and enterprises, showcasing their Agentic Smart City project in Vail, Colorado, which tackles 508 web‑accessibility remediation, deed‑restriction...