Frequently asked questions

Isn't this just a wrapper around an AI API?

No. The inference call is one part of a larger system. Eigin is a persistent agent with its own identity, knowledge graph, and context management that carries across every conversation. The model is a component — it's not the product.

How does the local/cloud split work?

The agent, its data, and its tools all run on device. If you're using a cloud provider, the inference call goes off device to that provider with your key. If you're using a local model, nothing leaves the device at all.

Why native apps rather than cross-platform or web?

Eigin does a lot of heavy lifting at the platform level - on-device machine learning, hardware-accelerated processing, background execution, deep OS integrations. Cross-platform frameworks have improved a lot, but there's a real cost in drift, performance overhead, and UI fidelity when you're working this close to the metal. Native is the best fit for what we're building.

Can I self-host the relay?

Yes. The relay ships as a Docker image and the repository includes Terraform code to deploy it. You can configure it yourself with your own API keys.

How does Eigin's agent model differ from frameworks like OpenClaw or NemoClaw?

Those are server-side, autonomy-first frameworks, designed to give an agent broad control over a machine or workflow. Eigin is edge-oriented: the agent runs on the user's device, serves the individual, and operates within a permission model that defaults to least privilege. It's personal AI, not an autonomous system.

How do I contribute?

See CONTRIBUTING.md.