Linzumi

What it does

Linzumi is a team chat platform where every channel can direct a fleet of AI coding agents. Agents run on your own machine, not in a hosted cloud, and every action is logged and visible in the thread. You can start a run, watch the diffs land, and steer agents mid-task — all from the same chat interface your team already uses. The platform supports parallel agents, inline diff review, and full audit trails.

Who it is for

Linzumi is built for engineering teams that want to use AI coding agents collaboratively. It targets teams that have outgrown single-agent terminal workflows and need shared visibility, mid-run steering, and the ability to run agents on their own hardware. The product is especially relevant for teams at startups or scale-ups where throughput is bottlenecked by attention, not headcount.

Why it matters

Current AI coding tools force a trade-off: either run agents on your machine but fly blind (no team visibility, no mid-run intervention), or use hosted agents that are easy to watch but don't run on your real environment. Linzumi solves both: agents run on your machine with your credentials and environment, while the team sees every command, diff, and test output in the chat. This means problems are caught early, trust is maintained, and one operator can steer multiple agents in parallel.

Launch signal

Linzumi is currently in beta for macOS (Apple Silicon). It is backed by Y Combinator, Matrix, SV Angel, Decibel, Pioneer Fund, and Axiom Partners. The website offers a free Personal tier and a $100/month Company plan with unlimited team members. An Enterprise plan is available for orgs with SSO, self-hosting, and custom guardrails.

Brand and naming

The name "Linzumi" suggests a blend of "link" and "sumi" (possibly from Japanese for "corner" or "ink"), evoking connection and the idea of bringing AI agents into the team's corner. The tagline "A chat where work runs wherever it gets unblocked" positions the product as a frictionless collaboration tool. The brand emphasizes trust, control, and parallel leverage — appealing to engineering leaders who want to scale AI usage without losing oversight.