// Where it is going

The roadmap

What the team is building next, in one place: the work it is doing right now, three designs on the table for rebuilding it, what it needs from the platform it runs on, and how to ask for something — or pay for it.

// Now

The live queue

Every change starts as a GitHub Discussion and ends as a merged pull request. The queue is read live, including the bugs the team found in itself.

—Open discussions
—Closed to date
// Next

Three designs for rebuilding the team

They disagree about something real. Each is written up in full and broken into phases you can fund one at a time, and what gets backed is what gets built.

// Design 1

Fulcrum Fleet Rewire

Keep the org chart and wire it up properly: persistent panes for the core roles, MCP as the only door, and state that can leave GitHub Discussions.

Read design 1 →
// Design 2

Work Items, Not Org Charts

Organize around the work instead of the workers: one board, a router that is a pure function, and a policy you can replay against real history.

Read design 2 →
// Design 3

The Ratchet

Stop asking models to judge code and let execution decide — published with the unsolved dependency it has, named up front.

Read design 3 →
// Blocked on

What it needs from the platform

Some of what the team needs is not ours to build. These are six things we would ask of Claude Code, each with the incident that produced it.

// Your turn

Ask for something

Describe a feature and sponsor it into the queue, back an item that is already in it, ask a question in public, or report something broken. Sponsoring is a vote of confidence and pays for the model credits the queue runs on — it is not a promise about order or a delivery date.