Install the Arango AI SDK

Experimental

Requirements

• Python 3.12 or newer. The package declares support for Python 3.12 and 3.13.
• uv , which manages the SDK’s environment, and Git to clone the repository.
• An Arango Contextual Data Platform deployment with the Agentic AI Suite, reachable from wherever your code runs. See the AutoGraph prerequisites for the platform and ArangoDB versions AutoGraph needs.
• A username and password for an account that can access the database you want to work in. The SDK signs in with these; it does not take a token.
• An API key for a model provider, for example OpenAI, for the chat and embedding models AutoGraph uses.

Install the package

The SDK is open source under the Apache License 2.0. It is not on the public PyPI yet, so you install it from a clone of the arango-ai-sdk repository  on GitHub:

git clone https://github.com/arangoml/arango-ai-sdk
cd arango-ai-sdk
make install

make install runs uv sync, which creates a virtual environment in .venv with the SDK and its dependencies. Without make, for example on a default Windows setup, run uv sync directly.

Activate the environment, then run any script that imports the SDK from that shell:

source .venv/bin/activate

.venv\Scripts\activate

Verify the installation

import arango_ai_sdk

print(arango_ai_sdk.__version__)

Next step

Continue with Getting started to build your first project.