Philosophy

"People talk about the most personal sh*t in their lives to ChatGPT"

- Sam Altman

He's right. People bring their plans, worries, health and relationships to personal AI. Today, all of it lands on someone else's servers, handed to a stranger every time they ask for help.

Personal AI is too personal to be anyone's but yours. Eigin is built on that conviction.

Large language models are becoming commodities. Choosing one should feel like choosing a phone carrier or a utility: pick one that works, and switch when you want. The utility can change. The house (and everything in it) stays yours.

Eigin is free and open source, because personal AI should belong to people.

Goals

These are what we're trying to achieve when we design Eigin.

Ownership

Eigin is local-first. Conversations, the agent, its memory and its tools live on the user's device. That puts users in control of their own data, which strengthens privacy, security and individual digital sovereignty.

Eigin works without an internet connection. The cloud is an upgrade, not a requirement.

Authorship

Eigin is malleable software. Users are the authors of their AI, not just consumers of it. The agent starts as a blank slate with sensible defaults, then takes shape around the person using it.

The user decides what Eigin is: its character, skills, capabilities, goals and more.

Human at the helm

Eigin exists to support humans. It isn't a replacement for humans, or a separate entity with a will of its own. People decide what it does, within limits they accept and limits that are good for people in general.

For everyone, every day

Eigin should be something people reach for daily, without being power users. There's real complexity in the AI stack. Our job is to absorb it, so owning a personal AI is accessible to anyone.

Non-goals

These are the things we actively steer away from when we design Eigin.

Not SaaS

Eigin is software for the user's device. We're about empowering the individual, not creating lock-in or dependency. Our added services are optional and self-hostable.

Not always on

Eigin isn't built to run unattended for long stretches. We don't pretend AI agents can do that reliably, so our focus is what can be done safely and reliably, every time.

Not a model zoo

Eigin isn't for benchmarking, tinkering or trying every single model. It's for getting things done.

Not everything

Eigin doesn't try to do everything. We'd rather do a few things well: the local-first agent, deeply personal context, and conversation through voice and text.