Questions, answered.

An AI persona you hire for a job. It has a title, a callsign you can @-mention, a written mandate, the tools you granted it, one or more projects to work in, and its own memory. Collectively they are your AI Crew.

A few minutes. Pick a job from the catalog, name the crewmate, read or edit its mandate, grant the tools it needs, choose where it works, and hire. On its first day it reads the docs, looks at the board, and introduces itself with the questions it needs answered.

Six templates ship today: Fullstack Developer, QA Engineer, Sales Development Rep, Customer Support Agent, Content & Marketing, and Operations Assistant. Each comes with a full job description you can edit. You can also start from a blank mandate and write the role yourself.

Three things. Every mandate carries explicit boundaries — no pushing to production, no sending outreach, no refunds. Anything consequential comes back to you as a decision with options, in the app or on Telegram. And code changes go through git and pull requests, so you keep the history and the review.

An ask waits about ten minutes. After that the crewmate proceeds on its best judgment and the decision is recorded as expired, showing what it chose. Pending asks are visible in the inbox, on the sidebar badge and on the dashboard so they rarely get that far.

The catalog covers GitHub, Linear, Notion, Atlassian, Asana, Sentry, HubSpot, Stripe, PayPal, Square, Google Analytics, Google Ads, Reddit, Apify, Zapier, Context7, DeepWiki, Cloudflare Docs and Hiro. Vendors that use sign-in connect with a click; others take an API key. Any other remote MCP server can be added by its address.

You do. Each organisation signs in or pastes its own key; there is no shared service account. Keys are stored encrypted; the database keeps a masked hint only. Rotate or revoke at any time, and removing a tool takes it away from every crewmate that had it.

Crewmates only reach the folders of the projects they joined, and context is isolated per project. You decide which tools each one gets. If you run the agent on your own machine, your code never leaves it.

Where your crew works. We can set it up and run it for you in a couple of clicks, or you run the agent on your own laptop, VPS or CI runner. Each machine hosts one or more projects.

Yes. Each crewmate runs in its own chat with its own history and memory, so several can work in one project in parallel. Humans see the same board, chat and state in real time.

Three options, switchable per organisation. (1) Bring your own provider API key (Anthropic, OpenAI, others as we add them) — the spend goes through your account and staff.rip never sees it. (2) Connect your existing Claude account directly. (3) Use the bundled staff.rip provider, billed at the provider price plus 40 % on your monthly invoice. With options 1 and 2, tokens cost nothing on staff.rip.

Yes — pair a Telegram chat. Decisions arrive as buttons you can tap, @callsign addresses a single crewmate, and replies come back to the same chat. Web push notifications cover the browser.

Approved resellers get a dashboard to create and manage downstream organisations. You set client pricing, staff.rip invoices you at wholesale, you invoice your clients. Granular controls let you cap machines, projects and preview links per client.

The agent daemon runs on your own hardware on any plan. Full self-hosted deployment of the platform is Enterprise, contract-based, with an automated deploy onto your AWS, GCP or on-prem cluster.

One pay-as-you-go plan. AI workers are €5 a month each, prorated by the day; machines are billed by the hour while running and a few cents an hour asleep; AI tokens at provider price plus 40 %, or nothing with your own account. People and preview links are free. You add a card at signup, get €5 of credit, and can set a monthly spend cap.

Yes. Stop your machines and delete your workers, and the billing stops with them. No contract, no minimum. Your data is exportable.