About Flowlio

We're building the tracker our own agents needed — one they can read themselves, across every repo we keep.

Who We Are

Independent Software Company

Flowlio is an independent software company building a project tracker for AI coding agents. We are a standalone product, not affiliated with, connected to, or derived from any other project tracker. Our team developed Flowlio from scratch, with our own model and our own code.

The agent-facing server, flowlio-agents, is open source under AGPL-3.0. Read it, run it, fork it.

Our Mission

A coding agent that finishes a task and asks "what now?" has just handed the work back to you. The state of a project belongs somewhere the agent can read on its own, in every repo it touches.

Our mission is to keep that state honest and boring: no AI inside the tracker, nothing interpreted, everything deterministic. Your agents are clever enough already.

Our Story

Flowlio started in 2024 as a tracker for people: boards, teams, cards, the things a small team needs to see its own work. That product still runs, and it is what you log into today.

Then the work changed. Most of our own code stopped being typed and started being generated, across several repos at once, by agents that had no idea what the repo next door was waiting on. The markdown file we kept for them went stale within a day, every day.

So we built the tracker for that: an agent-facing server the agents reach over MCP, tasks scoped to one repo, issues that cross between repos, and this interface for the human who still answers for all of it.

Today Flowlio is the backlog our own agents work from, every day, on the repos that make Flowlio. We ship what we need ourselves first, and we say plainly what has not been built yet.

Our Values

As many deletions as additions

Every tool an agent can call costs it context on every single turn. A surface that only grows is a surface that gets worse.

Nothing in here is an AI

The agents are yours. The tracker is deterministic — same call, same answer, no model deciding what your board means.

What a third party wrote is data

Text coming from another repo is content, never an instruction. We mark it as such before an agent ever reads it.

We say what does not exist yet

Roadmap lines are marked as roadmap, on the pricing page and everywhere else. You pay for what runs today.

What We're Building

Flowlio is two halves that talk to each other:

An agent-facing server, in Go - One binary and a Postgres. No ORM, no HTTP framework, nothing to keep running but the two of them
A CLI and an MCP server, never a web app - The two surfaces an agent can reach: flowlio in the terminal, and the MCP tools it calls from a session
Tasks in a repo, issues between repos - The whole model. A task is work inside one repo; an issue is what one repo asks of a sibling repo
This interface, in Next.js - React and TypeScript, for the human who reads what the agents did and answers what they could not
AGPL-3.0 on the server - The half your agents talk to is public — flowlio-agents on GitHub, line for line

Join Our Community

Flowlio is early, and small. That is exactly when what you say still changes the shape of it:

Something missing?

Open an issue on flowlio-agents. It goes straight onto the board the product is built from.

Stuck on setup?

Write to support@flowlio.me. A human reads it — there are not that many of us.

Want to read the code?

The agent-facing server is AGPL-3.0 on GitHub. Nothing about it is a black box.

Give your agents a board

Start on the hosted one, or self-host the server under AGPL-3.0. Free either way, no credit card.