Feature requests pile up faster than anyone reads them. Kai PM is the part of Gleap that reads them for you: it scores every request, asks customers for the details you'd have asked for, decides the obvious ones, and answers questions about your roadmap in chat.
It lives under Feedback portal, in the Kai PM section of the sidebar.
Scoring. Every open request gets a score from 0 to 100, built from how many people want it, how much revenue sits behind them, how well it fits your goals, and how much work it is. The factors are visible on every request, so the number always has a reason attached. See How Kai scores a feature request.
Intake. When a request arrives, Kai can ask the customer one or two clarifying questions before it enters your queue, merge it into an existing request when it's a clear duplicate, and tell the customer when the feature already exists. See What Kai does when a request arrives.
Triage. With Let Kai decide switched on, Kai accepts or declines each new request itself, replies to the customer in their language, and moves accepted requests onto the public roadmap where others can vote. You see every decision and can reverse it. See Let Kai accept or decline requests.
Chat. Ask Kai PM anything about your roadmap: what to ship next, which requests are duplicates, what the top accounts are asking for. It can also do things, like changing the status of fifty requests or merging duplicates, but only after showing you what it's about to do. See Ask Kai PM in chat.
Kai acts alone on the things that are cheap to undo: scores, clarifying questions, merges it's confident about, and, if you switch it on, accepting or declining new requests. Accepting is safe because it only publishes a request for voting; nothing gets built.
Building, releasing and messaging everyone who voted stay with you. Kai will prepare the message and the pull request, but a person clicks send.
Everything Kai does is listed on the Activity page, and every automatic decision shows on the request itself with Kai's reason.
Kai PM is included in the Pro plan and above. The overview and scoring work for everyone on the project; the chat is Pro-only.
Requests scored on revenue need the accounts behind them to carry a value. That's one field in your identify call, covered during setup.
Effort estimates get much better when a code repository is connected under AI & Automation → Knowledge. Without one, Kai estimates from the request text alone.
