​
Pre-requisites

Follow the installation instructions and run baml init in a new project.

The starting project structure will look something like this:

​
Overview

Before you call an LLM, ask yourself what kind of input or output youre expecting. If you want the LLM to generate text, then you probably want a string, but if you’re trying to get it to collect user details, you may want it to return a complex type like UserDetails.

Thinking this way can help you decompose large complex prompts into smaller, more measurable functions, and will also help you build more complex workflows and agents.

​
Extracting a resume from text

The best way to learn BAML is to run an example in our web playground — PromptFiddle.com.

But at a high-level, BAML is simple to use — prompts are built using Jinja syntax to make working with strings easier. We extended jinja to add type-support, and static analysis of your template variables, and a realtime side-by-side playground for VSCode.

We’ll write out an example from PromptFiddle here:

baml_src/main.baml
// Declare the Resume type we want the AI function to return
class Resume {
  name string
  education Education[] @description("Extract in the same order listed")
  skills string[] @description("Only include programming languages")
}

class Education {
  school string
  degree string
  year int
}

// Declare the function signature, with the prompt that will be used to make the AI function work
function ExtractResume(resume_text: string) -> Resume {
  // An LLM client we define elsewhere, with some parameters and our API key
  client GPT4Turbo

  // The prompt uses Jinja syntax
  prompt #"
    Parse the following resume and return a structured representation of the data in the schema below.

    Resume:
    ---
    {{ resume_text }}
    ---

    {# special macro to print the output instructions. #}
    {{ ctx.output_format }}

    JSON:
  "#
}

That’s it! If you use the VSCode extension, everytime you save this .baml file, it will convert this configuration file into a usable Python or TypeScript function in milliseconds, with full types.

All your types become Pydantic models in Python, or type definitions in Typescript (soon we’ll support generating Zod types).

​
2. Usage in Python or TypeScript

Our VSCode extension automatically generates a baml_client in your language of choice.

from baml_client import baml as b
# BAML types get converted to Pydantic models
from baml_client.baml_types import Resume
import asyncio

async def main():
    resume_text = """Jason Doe
Python, Rust
University of California, Berkeley, B.S.
in Computer Science, 2020
Also an expert in Tableau, SQL, and C++
"""

    # this function comes from the autogenerated "baml_client".
    # It calls the LLM you specified and handles the parsing.
    resume = await b.ExtractResume(resume_text)
    
    # Fully type-checked and validated!
    assert isinstance(resume, Resume)


if __name__ == "__main__":
  asyncio.run(main())

​
Further reading

Testing AI functions