Your users write in chats. Your work lives in code.
If you build software, the bug reports and wishes arrive on WhatsApp, Slack and email, and the answers are in your repository. ChatPilot connects the two: the AI reads your code to check what a user reports, the ticket goes where your work is tracked, and the reply goes back when it's fixed.
“Is this a real bug? Check the code.”
Give ChatPilot the folders it may read: your code, docs or notes. In a private discussion next to a chat, the AI looks things up there (read-only) before it answers you: what the user actually hit, where it comes from in the code, and a reply you can send. Paste a screenshot of the error into your question if that's quicker.
- New messages that look like a bug report or a request are flagged, with their own filter in the inbox
- The AI can only read, never change: no tool that writes is ever given to it
- What users write is data, not instructions: a message can't tell the AI what to do
From the chat to your tracker, and back
When the AI agrees it's a bug, it proposes a ticket on a card: title, description and why, filled in from the conversation. You choose where it goes and press Run:
- GitHub issues in a repository you add, signed in with GitHub (a short code, no token to paste)
- Your tracker or helpdesk as a connected system, such as Linear, Jira, Sentry or your own MCP server
- A webhook in Zapier, Make, n8n or your own endpoint, for everything else
Where the system can show it, the ticket is followed afterwards: when its status changes you hear it in Telegram, and in the chat it came from a reply to the person who reported it is ready for your Send.
The GitHub questions waiting on you, in your inbox
Teammates and contributors @-mention you on issues, or put a label on the ones that need your decision. ChatPilot collects those, with a count, under Questions for you and in your Telegram bot. Answer in the box, or press Go as proposed; from Telegram you see exactly what will be posted before you post it. Nothing is posted without your press.
Reported something to us on chat-pilot.ai? With a key from your account page, the questions we ask you about your reports show up in the same list, and your answer goes straight back to us.
Something wrong with ChatPilot itself?
Report a problem or a wish from the panel, from a discussion, or with /bug in your Telegram bot. You always see the full report before you press Send. Diagnostics are optional and shown in full first: the version, the system, the names of your connected apps and the last log lines, with phone numbers, e-mail addresses, keys and chat names replaced by placeholders, plus a few data checks that are counts only, so you never have to run a command to help with a bug. When we ask for one check by name on a question for you, Send data check under Questions for you runs it, shows you the result, and posts it only when you press Post. Never your messages, contacts or keys. Add up to five screenshots by pasting, dropping or picking them, or take one of the panel itself; each opens full size first so you can blur or black out names and messages, and it's sent exactly as you see it there, with the covered parts gone from the picture and its hidden details stripped. Signed in to your ChatPilot account in the app, you need no GitHub account and no key: your reports and our answers stay listed in the panel and under Your reports on chat-pilot.ai. The same sign-in brings your licence over with one press.
Make your own system work with ChatPilot
Any MCP server
Add your admin panel, shop or back office with its URL and a key, or sign in where the server supports it. ChatPilot learns what kind of system it is, what an account is there and which read tool gives what, and you can correct its reading.
Profiles and packs
A profile is a small data file (no code, no keys) that tells ChatPilot about a kind of system or one product: the client card fields, alerts, reports and example questions. Ship them as a pack, or let your MCP server offer its own profile. No MCP? A pack can describe a REST API as read tools instead, the way the Everflow, Affise and Scaleo packs do.
Runs where you do
macOS, Linux or Windows, on your desktop or on a server of your own (Docker included), reached from your phone over your own private network. Service, updates, backups and profiles also have their own commands.
Pick Builder / developer when ChatPilot asks what you use it for, and these tools come first. About the use modes.
Get an invite
ChatPilot is in a private beta for macOS, Linux and Windows. Leave your email and we'll send you an invite and the setup guide when there's room.
We only email you about your invite and the beta. Unsubscribe at any time. How we handle your email. This form is protected by reCAPTCHA: Google's Privacy Policy and Terms apply.