Skip to main content

Using AI assistance

Ask AI helps you understand accessible work and prepare changes for review. Useful questions include:

  • What needs attention in my work?
  • Which team can help with this issue?
  • What does this person do?
  • What is blocking this project?
  • Review this work item or entry.

AI can use project, team, user, and work-item context. It fetches more detail when the question needs it and only uses records you are allowed to access.

Project-specific questions work best with a project added as context. Ask AI can summarize the project graph, surface blocked or overdue work, explain decisions, and point to the records it used.

Dictate a message​

When voice transcription is available, select the microphone control beside Send in the Home or Ask AI composer and speak. The browser asks for microphone permission the first time you use it. While you record, a live waveform and timer show the audio being captured. Stop the recording to transcribe it, then review or edit the text before sending. Voice recordings are not saved as chat attachments.

Windrunner supports OpenAI transcription with gpt-transcribe and Gemini transcription with gemini-2.5-flash. Voice transcription is configured independently from the chat AI provider. See Voice transcription configuration to enable a provider.

Review changes​

AI may propose new or updated work items, entries, and relationships. It checks for likely duplicates first. Clear matches are reported for update; ambiguous matches require clarification.

Proposals are not applied automatically. Review the proposed fields and then accept, reject, or ask for a revision. Users, teams, and projects are not directly changed by chat proposals.

When a proposal targets project work, it also appears in the workspace as an AI proposed row. Use the accept or reject control on that row after checking the action, type, title, priority, status, parent, assignee, and due date.

Ask AI showing a proposed high-priority task beside the project workspace
Example: Ask AI drafts a high-priority task and marks it as not applied until someone reviews it.

Work item and entry AI Review actions follow the same review-before-save model.

Work item inspector showing an AI suggestion with suggested due date and blocker
AI Review shows field-by-field changes and suggested blockers before you apply the draft.

Measure AI usage​

Admins and superadmins can open AI Analytics to filter by project and time range, review request and token totals, inspect feature reliability, and track accepted changes and acceptance rate. The overview separates usage from outcomes so a high request count is not mistaken for successful automation.

AI Analytics overview with request, token, acceptance, and feature usage metrics
AI Analytics reports request volume, tokens, failures, and review outcomes for administrators.