Your AI agents build. You stay in control.

The problem

Writing the code stopped being the hard part

AI got good at building things, and it got cheap. Once it is working on more than one project, the hard part is everything you now have to do around it.

1. Tell me what to work on next, and why, without me rebuilding the whole picture first.
2. Let work continue while I'm away, but stop and ask when it hits a decision that is actually mine.
3. Show me evidence the work is really done, not an AI's summary of itself.

How to manage multiple AI coding agents without losing track →

Is Arcadia for you?

For the person responsible for AI-built work

You do not need to be a professional engineer. You do need to have AI building something real already — a product, internal tool, or software-backed business — and be the person who has to decide what happens next and whether the result is trustworthy.

• Engineers running more than one agent or project

  Keep dependencies, priorities, and completion evidence outside the agent tabs.

• Founders and business owners directing AI-built software

  Use a durable record to turn an idea already in motion into bounded work, clear decisions, and proof.

• Not an idea-to-product shortcut

  If you only have a raw idea, or want a better code-generating agent, Arcadia is not the tool yet. It governs an AI coding agent; it does not replace one.

How Arcadia uses specs and dependency graphs to manage AI coding work →Do you need to be an engineer? →Fit and availability, answered →

The approach

A layer above the agents

Mission Control does not write code. It decides what the things writing your code should be doing, and checks what they hand back.

One person can run many projects at once this way: Arcadia watches the work across all of them and brings you what actually needs your judgment, so you can oversee everything calmly without living inside any single one.

• Plans, actions and a work pointer

  Governed documents with one active plan and one current action.

  Working

• Decisions and approval gates

  Consequential calls are recorded and wait for a human answer.

  Working

• Completion evidence

  Acceptance criteria are checked before an action can be marked done.

  Working

• CI and review-finding repair

  A failing check or a reviewer's finding gets fixed and re-validated automatically, up to a bounded number of attempts, before it reaches you.

  Working

What governed autonomous development means →

How it works

Intent becomes verified software

One loop, running continuously. Each step is inspectable, and each step is somewhere you can intervene.

How sessions stay out of each other’s way →

Your agents

It does not replace Claude Code, Codex, or OpenCode

Bring your coding agent or harness. Arcadia provides the direction and checks; the coding system does the building. Arcadia currently drives Claude Code, Codex, and OpenCode. Coding agents remain interchangeable execution tools under your direction.

How this differs from orchestration frameworks and issue trackers →

What it is built to do

Five sessions, a week away, a vague idea

Scenarios Mission Control is designed for — imagined situations, not customer stories.

• Five sessions, three repositories, one Tuesday

  Each session knows its project, its action, and the branch it may touch.

• A week away, with the work still moving

  What needed your judgment waits in one list, with each choice’s consequence stated.

• A vague idea, by Friday, as something executable

  “The onboarding is confusing” becomes a plan with actions and acceptance criteria.

Where it actually is

What works today, and what does not

Arcadia governs its own development — these tags describe the last reviewed snapshot.

• Plans, actions and a work pointer: Working
• Arcadia Field Notes opens Canyon Pass: Working
• Earlier warnings for confusing run instructions: In progress
• Agent session management: Working
• Decisions and approval gates: Working
• Completion evidence: Working
• CI and review-finding repair: Working
• Multi-project portfolio view: In progress
• Install on someone else's machine: In progress
• Hosted Mission Control: Exploring
• Cost-aware model selection: Planned
• Idempotent Ask intake: Working
• Next-action clarification: Working
• Independent pull-request review: Working
• Peer-watch building blocks: In progress
• One to-do list for what needs your attention: Working
• A check that the next Action is ready to start: Working
• Free-text input coverage: Planned
• Standard harness protocol: Planned
• Workflow registration and projections: Deferred

11 Working4 In progress1 Exploring3 Planned1 Deferred

The full status board, with what each one means → Recent achievements and changes →

How we decide

Three rules that shape what gets built here

• Find the 20% that matters

  Do the small part that carries most of the value first.

• Don't build what you don't need yet

  A feature for a need that might show up later is debt today.

• If not now, then when?

  “Not yet” is a real answer only when it names what would change it.

Each rule, with what it changed on this site →

Early access

Help work out what this should be

Early users get direct access to the person building it and real influence over what comes next. We ask for honest feedback. Nothing is for sale, and there is no price yet. Nobody has paid for Arcadia; early-user feedback helps shape that decision.

Arcadia is real software in active development. It is not finished, it has no customers or measured productivity outcomes yet, and early users help decide what it becomes.