BAML - A New Programming Language for Building Agents
summary
BAML is a new domain-specific language for agent code that treats prompts as typed functions, provides built-in test cases, visualizes raw model requests and retry policies, and strips common hallucination artifacts into clean JSON. It ships as an airgapped and internet-free runtime, implemented in a polyglot Rust runtime with type-safe bindings for TypeScript, Python, Go, etc., claims to work with any model without fine-tuning (even GPT‑5), and reports adoption by 1,000+ organizations including Fortune 50s and government teams.
AI-written summary · about
chapters (10)
- 0:00AI model reliability today
- 1:23Introducing BAML
- 3:08Reliability shift for AI apps
- 5:14String prompt problems
- 6:38Prompts as functions in BAML
- 8:47Debugging BAML prompts
- 10:09Reasoning and type safety in prompts
- 11:52Reliable maintainable code bases
- 12:56BAML ease and adoption
- 13:56Future of BAML and agents
related talks
19mBAML: a new programming language for AI
50mLet's build an AI agent
36mProduction-Ready AI Agents: From LLMs to Small Language Models
01h 43mBuilding LLM Agents (Hackathon)
49mLet's build an AI agent
41mThree agents, three frameworks, one talk
49mAI Apocalypse Survival Kit: coding agents & Tiny Language Models
10mAI agents on the JVM