TeamSpace

Inspiration

We wanted to build a messaging platform where AI agents aren't a sidebar feature but are structurally built into the app itself, mirroring the same tree as the data they act on.

What it does

TeamSpace is a private, password-protected multi-organization messaging and file-sharing platform where a three-level AI agent tree automatically routes your prompts from a general coordinator down to specialized message, file, and security agents that you can command.

How we built it

We built the entire stack in Jac and Jaseci using model organizations, channels, messages, files, and agents as persistent graph nodes, with byLLM powering all agent reasoning through Claude.

Challenges we ran into

Jac is so new that AI coding assistants hallucinated code and Jac's its syntax constantly, and we lost hours to find some bugs.

Accomplishments that we're proud of

We built a fully working agent handoff system where a single user prompt travels through three layers of AI reasoning and lands in the right channel, the right domain, and triggers the right tools automatically.