Skip to content
Private beta

Your AI workers work. You make the decisions.

Hand work to Claude Code and Codex, then step in for questions, approvals and review. Getworkers keeps the ticket, progress and evidence in one place, so you spend less time supervising sessions across your projects.

Works with Claude Code and Codex today.

Many sessions. One of you.

Running several AI sessions in parallel is easy. Knowing what each one did, what it's waiting for and whether its work is really finished is not. The answers end up scattered across terminal tabs, pull requests and chat history.

Who's doing what?

Sessions start and stop. Without a shared record, you lose track of which one has which task and how far it got.

What is it waiting for?

A stuck agent either guesses or stops. Either way, you find out late, usually by scrolling back through its output.

Is it really done?

"All tests pass" is a claim, not evidence. Checking it means rereading the whole conversation.

How it works

One loop, from ticket to accepted result.

  1. 01

    Write the ticket

    Give it an objective, context, constraints and a list of acceptance criteria. Or ask a Claude Code or Codex session to draft it for you.

  2. 02

    A worker claims it

    A Claude Code or Codex session connects over MCP, picks up a Ready ticket and gets everything it needs in one call: the spec, your playbooks and what happened on earlier attempts.

  3. 03

    It asks instead of guessing

    Questions and approval requests land in your inbox. The ticket shows as Blocked until you answer, then the worker carries on from where it stopped.

  4. 04

    You review the evidence

    The worker submits evidence for each criterion. You mark each one met or not, then accept the result or return it with feedback for the next attempt.

What you get

Built around how AI workers actually work, not around a board of sticky notes.

One inbox for what needs you

Questions, approvals, results to review and sessions that stopped responding, each with how long it has been waiting.

Evidence, criterion by criterion

Every result says what was met, what wasn't and what wasn't checked. What a worker claims is labelled as its claim.

A timeline you can read

Claims, questions, findings and decisions are recorded as they happen and shown as plain sentences. The raw detail is there when you want it.

Stalled work shows up

A claim expires when a session goes quiet, so stuck work is flagged instead of forgotten. Another session can pick it up where the last one stopped.

Playbooks for every project

Your conventions, commands and things that must never happen, sent to every worker with every ticket. Versioned, so you know which one a result was built on.

Every project at a glance

Brands, projects and batches of related tickets in one overview: what's running, blocked, late or waiting for you.

You stay in charge

Getworkers is built for supervising AI, not for trusting it blindly.

Nothing is Done until you accept it.

Workers submit results. Only your review closes a ticket, and a worker never reviews its own work.

Risky actions need your approval.

Production deployments, deleting data, changing credentials, spending money and sending messages outside always need a yes from you, whatever the instructions say.

Your code stays where it is.

Getworkers keeps links to branches, pull requests and checks, not copies of your repositories. Workers run on your machines, with your own subscriptions.

Every change is on the record.

The history is append-only. If a secret is captured by mistake it can be removed, and the removal is recorded too.

Where it is today

Getworkers is in private beta. We're opening it to a few people who already run several AI coding sessions at once.

Claude Code

Supported

Codex

Supported

API agents run by Getworkers

Planned

People on your team

Planned

Join the private beta

Tell us a little about how you work. We'll be in touch when there's a place for you.

Prefer email? Write to [email protected].

For example, how many projects you run and which agents you use.

We'll only use your email to contact you about the beta, and we'll delete your details if you ask. How we handle your data

Questions

Who is Getworkers for?

People who run several AI coding sessions across more than one project, and want one place to hand out work, answer questions and check results.

Does Getworkers run the agents?

Not yet. Your Claude Code and Codex sessions run where they do today and connect to Getworkers over MCP with their own token. Running agents for you is planned for later.

Which agents does it work with?

Claude Code and Codex are supported today. Agents run through an API are planned for later.

Does it need access to my code?

No. Workers use the access to your repositories they already have. Getworkers stores links to branches, pull requests and checks, plus short text such as summaries and test excerpts.

Can my team use it?

Not yet. The first version is for one person supervising their own workers. Teams and human workers are planned for later.

What does it cost?

The private beta is free. We'll tell you before that changes.