See what an AI agent will share, and report anything that goes wrong
Every place in Pekan where an AI agent generates content now says so, names the provider, and gives you a plain-language line about what that run can see, right in the launch dialog, before you confirm, not buried behind a details toggle you have to know to open. Breaking a task down, drafting a project plan and refining a Triage capture all show the same disclosure, and the review screens for those proposals now carry a clear AI-generated marker while you edit them, matching what task cards and comments already showed.
Alongside that, every surface that shows AI-generated content, task descriptions, comments, handoffs, breakdown and plan proposals, refined Triage items, now has a discoverable Report action next to it, plus a general one from Help. Reporting opens an email addressed to Pekan support with the feature and provider already filled in and space to describe what went wrong; the generated text itself is only attached if you tick a box to include it, so nothing you did not choose to share ever leaves your machine.
Column-hook automations that run an agent whenever a task changes columns fire the same way they always have, but the warning shown while you configure one now says plainly which providers it can start and that no one gets asked to confirm each time it fires.
Highlights
- Every AI run shows its provider and what it can see before you confirm.
- A Report action sits next to every piece of AI-generated content, plus a general one in Help.
- Reports never include the generated text or your project data unless you choose to attach it.