Skip to main content
BAML is a domain-specific language designed to make prompt engineering easy by turning it into schema engineering.

​
What is BAML?

BAML (Basically a Made-up Language) is a simple prompting language for building reliable AI workflows and agents. Instead of wrestling with f-strings and JSON schemas, BAML lets you focus on what matters: the structure and logic of your prompts. You don’t need to write your whole app in BAML - only the prompts! BAML generates type-safe client code that works seamlessly with your existing Python, TypeScript, Ruby, Go, or REST applications.

​
Core Philosophy

The fundamental principle in BAML is simple: LLM Prompts are functions. Every prompt is a function that takes in parameters and returns a type. 
function ExtractResume(resume_text: string) -> Resume {
  client "openai/gpt-4o"
  
  prompt #"
    Parse the following resume and return structured data.
    
    {{ resume_text }}
    
    {{ ctx.output_format }}
  "#
}

class Resume {
  name string
  education Education[]
  skills string[]
}

class Education {
  school string
  degree string
  year int
}

​
Key Features

Test 10x Faster

Built-in VSCode playground lets you test prompts directly in your editor with hot-reloading, no manual JSON schemas needed.

Fully Type-Safe

Get autocomplete and type safety in Python (Pydantic) and TypeScript, even when streaming structured data.

Works with Any Model

Supports OpenAI, Anthropic, Gemini, Vertex, Bedrock, Azure, and anything OpenAI-compatible (Ollama, OpenRouter, VLLM, etc.).

Reliable Tool Calling

Works even when models don’t support native tool-calling APIs using our SAP (schema-aligned parsing) algorithm.

Easy Model Switching

Switch between 100s of models in just a couple lines. Built-in retry policies, fallbacks, and round-robin.

Beautiful Streaming

Generate streaming UIs with fully type-safe partial responses. Includes React hooks for NextJS.

​
Use BAML from Any Language

BAML generates a baml_client that you can import into your application: 
from baml_client import b
from baml_client.types import Resume

resume = b.ExtractResume(resume_text)
print(resume.name)
print(resume.education[0].school)

​
Built for Production

100% Open Source

Apache 2.0 license. Built in Rust for maximum performance.

100% Private

No network requests beyond your explicit model calls. Not used for training data.

Version with Git

BAML files live in your codebase and can be checked into Git for easy diffs.

​
Why BAML?

Just as JSX/TSX revolutionized web development by providing the right abstraction for UI, BAML provides the perfect abstraction for prompt engineering:
  • No more f-string hell: Write structured prompts, not string concatenations
  • Schema engineering over prompt engineering: Define your output types, let BAML handle the prompt formatting
  • Instant iteration: Test changes in seconds, not minutes
  • Production-ready: Used by many companies in production with weekly updates
Try BAML online at Prompt Fiddle or explore Interactive Examples

​
Next Steps

Quick Start

Get BAML running in under 5 minutes

Why BAML?

Learn why you should use BAML over other frameworks

Build docs developers (and LLMs) love

Get started for freeTalk to us