Open, self-hostable infrastructure for reliable AI agents.
We build the infrastructure between AI models and production systems. Our projects prioritize user-owned data, portable interfaces, open protocols, operational durability, and claims backed by reproducible evidence.
| Project | Purpose | Maturity | Start here |
|---|---|---|---|
| mem | A portable, self-hosted memory plane for AI agents, with one core across API, MCP, CLI, and UI. | Experimental, active development | Product spec · Run locally |
| doc | A portable, self-hosted document plane for people and AI agents, with collaborative editing and version recovery. | Experimental, active development | Product spec · Run locally |
| digital-employee | A self-hosted runtime for role-based digital employees built from approved knowledge sources and tools. | Early development | Repository and current status |
Detailed capabilities, limitations, setup instructions, and roadmaps live in each project's repository. The maturity labels above describe the current public state; they are not compatibility or production-readiness guarantees.
- Start with a documented problem or outcome.
- Make changes through issues and pull requests, with passing checks and independent review.
- Validate claims with reproducible evidence.
- Prefer small, reviewable changes and transparent decisions.
- Keep shipped capabilities, experimental work, and future direction clearly separated.
- Choose the project closest to your goal and read its status, setup, and contribution documentation.
- Use the repository's issue templates for bugs, proposals, documentation, maintenance, and questions.
- Report vulnerabilities privately through the affected repository's Security tab.
Organization-wide policies:
A repository's local policy supplies project-specific instructions while preserving the organization baseline.