The assistant you already have open can now work on your beta program directly. Not a summary you paste into it, not an export you upload: your betas, testers, tickets and results, reached from the conversation.
Features
Stomio works inside your AI assistant
Stomio is now a remote MCP server. Connect claude.ai, Claude Desktop, Claude Code, ChatGPT or any other assistant that supports remote MCP, and it can work with your workspace directly.
More than eighty tools are available, covering beta projects, tasks and steps, tickets, testers, announcements and applicant screening. Ask it what your testers reported this week and which tasks they are dropping out of. Ask it to draft the week-two task set from the program you ran on the last version, triage the tickets that came in overnight, or write the mid-beta announcement and send it to the testers who finished week one.
Connecting takes about a minute: copy the connect URL from Workspace settings, then AI tools, paste it into your assistant, and sign in. The consent screen is where you choose how far the connection goes, reading only or reading and doing, and that choice is fixed for the connection, so widening it later means approving the wider access again. Every connected assistant is listed on the same settings page with what it was allowed to do and a Disconnect button.
Three things stay true whatever you or your assistant intend. Anything that would reach a tester or an applicant is previewed in the conversation and does nothing until you confirm it. The connection carries your own permissions and no more. And it only reaches beta projects that have AI features enabled, so a program can stay out of it entirely.
Our new how-to, Connect Your AI Assistant to Stomio, walks through the whole thing, and the AI page covers how this fits with the AI already inside the product.
Improvements
- Rate and Ranking steps can be edited on a running beta. Both used to lock the moment a beta started running, so a scale you wanted to widen, or a ranking option with a typo in it, stayed that way for the rest of the program. Because some testers may already have answered, Stomio tells you what the change means before you commit to it, and shrinking a rate scale warns you that responses above the new maximum will no longer appear in your results.
- Announcements can be twice as long. The limit went from 300 to 600 characters, which is enough for a mid-beta update that explains itself instead of a message that has to be split in two.
- The app updates itself quietly. The banner asking you to reload for a new version is gone. Stomio now picks up new releases on its own as you move around the app, so you stay on the current version without being interrupted to ask for it.
Fixes
- Character limits now count the characters you typed. Text containing an ampersand, a quotation mark or an angle bracket used to count against the limit several characters at a time, so a message could hit the cap well before it looked full.
- AI Canvas no longer fails outright on betas with a lot of history. It sizes its context to the model and works through them.
- Rate answers left over from a wider scale no longer distort the task detail breakdown.