Docs

Get started in 60 seconds.

The full technical docs live on GitHub. Here's the shortest path from zero to your first graduated rule.

bash

pip install gradata

Requires Python 3.10+

python

from gradata import Gradata

brain = Gradata(profile="my-agent")

# Feed it your first correction
brain.correct(
    draft="Sure thing! Here's the TPS report.",
    final="Here is the TPS report.",
    task="email_reply",
)

# Inspect what the brain learned
print(brain.rules(task="email_reply"))

Full documentation

Complete API reference, architecture deep-dives, and integration guides for Claude, GPT, Gemini, and local models are on GitHub.

Read the docs on GitHub →