DeepFounder AI

DeepFounder AI is an independent software laboratory, established in 2024, that builds AI agents and the infrastructure that makes them dependable in production: memory, tooling, coordination and storage. Its work spans agent development for real workflows such as customer operations, internal automation and knowledge retrieval, grounded in research into how agents behave outside a demo.

The lab also builds the plain interfaces agents run on, including REST, MCP and CLI, built in Rust and Python and released as open source. Deployment favors self-hosting on a laptop, workstation or private server by default, so data does not leave the building unless a user chooses to send it elsewhere.

Its published projects include desigma, a free, local-first design tool written in Rust with an MCP server for agents; castor, a self-hosted AI agent for business workflows; mnemos, a persistent cloud memory system for agents; and LineAgent, an issue tracker built for AI agents over REST, MCP and CLI. DeepFounder AI reports 8 open repositories, 5 systems written in Rust, 5 Android apps and more than 670 GitHub stars across its projects.

Beyond agent tooling, DeepFounder AI also publishes Android tablet apps: YoDraw for drawing and painting, AnMotion for frame-by-frame animation, Wemuse for turning handwriting into engraved sheet music, MIDIPad for turning a tablet into a USB MIDI controller, and SynthB, a two-layer subtractive synthesizer. All of the lab's code, interfaces and methods are published openly rather than kept secret.

Use cases

• Deploying a self-hosted agent for customer ops and automation
• Drawing with a full brush library on an Android tablet
• Turning handwriting into engraved sheet music