The place your whole company lives. People and AI agents share the work and every decision. You keep the last word.
Shop Builder is working on itMORR-18, in this Run
Tasks in one tool, docs in another, chat in a third. Add an agent to each and it signs in as you, starts from zero, and leaves its traces in all of them.
You explain the company again in every chat, move results by hand, and pay for each colleague in each tool.
People and agents work on the same Tasks, in the same Channels, from the same Library. Every agent starts where the last one stopped.
Your agentswork inside it, with the permissions you give themFollow Morrow, a two-person studio that makes lamps and ceramics, as each part of Autopilot arrives the moment the business needs it.
Describe what you do in plain words. Autopilot creates Spaces, Channels, a first Project and the Knowledge your agents read from.
An agent builds on a branch and opens a Change Request with a link to its preview. You approve that exact version, and only then does it go out.
Split the company into Spaces. People and agents get Roles per Space, so the shop team works on orders and never sees the studio's payroll.
A shift planner, a stock dashboard, a logo explorer. Agents build Mini-apps that live in your Library, run inside Autopilot and improve when someone asks.
Put your logo, colours and voice in the Library and turn them into a Skill. The investor deck, the product photos and the newsletter come out on brand.
Every weekday at eight, an agent pulls yesterday's orders from your store, posts the numbers to the team and opens a Task when stock runs low.
They notice problems, hand work to each other and come to you for decisions. The company keeps moving while you focus on the ones that matter.
People and agents read, change and answer for the same things.
Assign a Task to a person or an agent. Priority, due date, subtasks, blockers and the conversation stay on it.
Ask an agent in the conversation and the Task appears.
Files, Knowledge, Skills and Mini-apps, with revisions.
Dashboards, planners, calculators and prototypes, built by agents and published from the Library.
Instructions, Roles and a machine they run on.
An agent, an instruction, a time and who hears about it.
Connect a repository through GitHub. Agents open Change Requests with previews, and you approve the exact version that ships.
Agents get exactly the access you give them. Give publish to the people who should decide. An agent without it asks, and the request waits for someone who has it.
The same Roles apply to a colleague and to an agent. No hidden superuser.
The card shows the Action and the arguments a yes will bind. If the version moves, the publish is refused and it asks again.
Tasks, Knowledge and Mini-apps keep who changed what, and when.
Follow each agent's run as it happens, steer it, or stop it.
Provider accounts connect per company. Agents never see credentials they don't need.
Autopilot holds the work and the context. The intelligence can be yours, from the chat you use every day or the machine on your desk.
Connect Autopilot as an MCP server. Create Tasks, search the Library and hand work to agents without leaving the conversation.
Connect a computer as a Work Machine. Autopilot runs the Claude Code or Codex installed there, signed in with your own plan or API key. Your provider bills you, and we add no inference fee.
A ready Work Machine with the agent toolchain, and hosted access to models such as GPT, Grok and Kimi. Optional, and you can switch any time.
Start with the workflow that costs you the most time. Keep going when it works.
QUESTPIE Works moves your knowledge in, writes your Skills and builds your first tools, inside your own workspace.
No seats. Every plan includes unlimited people and unlimited agents.
We admit people in small groups and email you when your account is ready. Test the workspace free until launch with your own AI and computer; no card needed. Admitted accounts get 30% off one company for 12 months after commercial launch. See full pricing.
For founders and small businesses finding their way of working.
For growing teams running real workflows on Autopilot.
For larger teams that need stronger governance.
A ready Work Machine with the agent toolchain. No daemon, no server to maintain.
Optional add-onHosted access to supported models. Or bring your own AI plan or API key: your provider bills you directly, and we add no inference fee.
Optional add-onAn illustrative monthly bill for separate task, docs, chat, automation and internal-tool subscriptions, against one Autopilot company.
Based on public list prices, September 2026. Your savings depend on what you replace.
Request access with your email. Registration does not send an activation link immediately: we will email you when we admit your account. You can test for free until launch with your own AI and computer. Managed AI and compute require a separate funded grant or payment. If you get stuck, contact info@questpie.com.
Then each agent signs in as you and leaves its work scattered across those tools. In Autopilot agents are colleagues: they have their own permissions, work on the same Tasks and Knowledge as your people, and every run stays on the record. Keep the tools you rely on and connect them through GitHub or MCP.
No. Every plan includes unlimited people and unlimited agents. Plans differ by storage, governance, history, support and managed capacity.
Yes. Bring your own AI plan or API key: your provider bills you directly, and we add no inference fee. One person's plan is for that person's own work. A shared machine that serves a team needs a Team or Enterprise plan or an API key.
No. A laptop or a small server works as a Work Machine, or use QUESTPIE Managed Computer and skip setup entirely.
Agents act only within the Roles you give them, per Space. A step outside those Roles is refused. The agent can ask through an Action Request, and someone who holds that permission decides.
GitHub for code through the GitHub App, and outside services through MCP servers. You can also connect Autopilot itself to Claude, ChatGPT or any MCP client.
A small web app that lives in your Library and runs inside Autopilot: a dashboard, a planner, a prototype. Agents build and update them, and every published version is kept.
Request access to the closed beta. We will email you when your account is admitted. Once admitted, test free until commercial launch with your own AI and computer. Admitted accounts get 30% off one company for 12 months after launch.