Subvolts take Make AI programmable: learn about DSPy, a new way to control LLMs beyond simple prompts. Nov 18, 2025 - Developers
- Learn about DSPy, a new framework for programming LLMs that aims to improve control and reliability.
- Business
- Understand how new tools can make AI systems more predictable and useful for business applications.
This episode features a researcher presenting a novel approach to LLM programming that could significantly impact AI development tools.
How Foundation Models Evolved: A PhD Journey Through AI's Breakthrough Era
Audio streams from a16z. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
This AI + a16z podcast episode features Omar Khattab discussing his journey creating DSPy, a new programming paradigm for LLMs. Khattab argues that current approaches to interacting with AI are backward, proposing a middle ground between code and natural language to make AI systems more programmable and reliable. The conversation explores the implications for AI's future as infrastructure.
Imagine telling a computer exactly what you want it to do, but using a mix of simple instructions and plain English. This episode explains a new method called DSPy that makes AI models more predictable and easier to program, moving past just typing questions.
This episode is for developers and researchers grappling with LLM control. Khattab's DSPy offers a concrete solution to the "prompt engineering" problem, aiming to make AI intent programmable. This directly addresses a core challenge for all frontier labs like OpenAI, Google Gemini, and Anthropic as they build more capable models. Listeners will learn about a potential new layer for AI development, moving beyond simple prompts to more structured interaction.
- Hosta16z
- GuestOmar KhattabStanford PhD, Stanford
- Guest: Omar Khattab, Stanford PhD
- Show: AI + a16z
- Length: 57 minutes
- Topic: DSPy, a new paradigm for making LLMs programmable
- Problem Addressed: The gap between user intent and AI instruction
The Stanford PhD who built DSPy thought he was just creating better prompts—until he realized he'd accidentally invented a new paradigm that makes LLMs actually programmable.
While everyone obsesses over whether LLMs will get us to AGI, Omar Khattab is solving a more urgent problem: the gap between what you want AI to do and your ability to tell it, the absence of a real programming language for intent. He argues the entire field has been approaching this backwards, treating natural language prompts as the interface when we actually need something between imperative code and pure English, and the implications could determine whether AI systems remain unpredictable black boxes or become the reliable infrastructure layer everyone's betting on.
Follow Omar Khattab on X: https://x.com/lateinteraction
Follow Martin Casado on X: https://x.com/martin_casado
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
- Foundation Models
- Omar Khattab
- The large, general-purpose AI models that power many AI applications.
- LLMs
- Omar Khattab
- Large Language Models, AI systems trained on vast amounts of text data.
- AGI
- Omar Khattab
- Artificial General Intelligence, AI with human-like cognitive abilities.
- Paradigm
- Omar Khattab
- A fundamental way of thinking about or doing something.
24 minHow AI Is Rewriting the Rules of Cybersecurity | Truffle Security & Socket
WithDylan Ayrey, Feross Aboukhadijeh
47 minInside vLLM: The Engine Powering Open-Source AI
WithSimon Mo
41 minBuilding Self-Accelerating AI with Mirendil
WithBehnam Neyshabur, Harsh Mehta
43 minIdeogram’s Open-Weights Image Model and the Future of AI Design
WithMohammad Norouzi
1 h 13He Co-Invented the Transformer. Now: Continuous Thought Machines - Llion Jones and Luke Darlow [Sakana AI]
WithLlion Jones, Luke Darlow
1 h 01
1 h 42AI This Week: NVIDIA’s Record Revenue, Elon’s Data Centers in Space & Gemini 3’s Insane Performance w/ Salim Ismail, Dave Blundin & Alexander Wissner-Gross | EP #210
WithSalim Ismail, Dave Blundin, Alexander Wissner-Gross
1 h 07Google Pushes OpenAI, Bezos Returns, AI’s No. 1 Hit
WithRanjan Roy
Questions people ask
- What is DSPy?
- DSPy is a new programming paradigm that makes large language models (LLMs) more programmable and controllable, moving beyond simple natural language prompts.
- What problem does DSPy solve?
- It addresses the gap between what users want AI to do and their ability to instruct it, aiming to make AI systems more predictable and reliable.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to a16z.