Public README

David was the first entity. What began as a single GPT with a custom system prompt grew into a full AI orchestration platform.

Project David is an open source AI orchestration substrate — a platform for building, running, and coordinating intelligent agents at scale.

Built for engineers who won't trade data sovereignty for convenience. Run open-weight models. Own your infrastructure. Orchestrate at scale.

Project David is structured as a layered ecosystem. The platform is the core — a FastAPI orchestration engine that manages agents, tools, threads, and state. The SDK gives developers a clean Python interface to the platform without needing to interact with the REST API directly. david-core handles cloud deployment, TLS, and routing for the hosted instance.

The orchestration layer supports multi-agent deep research out of the box — a supervisor agent plans and delegates, workers execute in parallel, and a shared scratchpad provides observable, attributed state across the entire run.

Run inference and fine-tuning on your own GPU hardware. Scale horizontally across nodes or vertically within them — from a single home lab machine to a full enterprise cluster.

Full documentation is available in the docs repository.

When the hosted docs site is live, all links above will be updated to docs.projectdavid.co.uk.

Install the SDK:

pip install projectdavid

Run your first inference:

from projectdavid import Entity

client = Entity(api_key="your_api_key")

assistant = client.assistants.create_assistant(
    name="my_assistant",
    instructions="You are a helpful AI assistant.",
)

thread = client.threads.create_thread()

See the Quick Start guide for the full example.