Working with the AI assistant
The AI in killBottleneck is a drafting partner for your maps. One rule governs all of it: the AI proposes, you confirm. Nothing is written into your map without you seeing the change first — and the last AI change can always be undone with one click.
Have a map drafted
Next to the New project button sits With AI → Suggest with AI: describe your goal in a sentence or two, pick the map scope — Brief (main pillars) / Detailed (medium) / In-depth (thorough) — and run it. The AI sometimes asks a few follow-up questions for a better structure; the result arrives as a preview, and only Accept creates the goals. The confirmation reads "Structure added — N goals were added to the map."
Refine it in the chat
Inside the map editor, AI chat works over the map you're looking at. When you don't know what to type, the quick buttons help: Suggest changes, What am I missing?, Simplify the map. The reply arrives as a Suggested changes (N) card with Add / Edit / Delete rows and Apply / Discard buttons — every operation runs only after your confirmation. Undo for the last AI change sits right next to it.
From text or a voice memo
Map from text (also under "With AI") takes your notes — Upload .txt / .md, pasted text, or Upload audio, which is transcribed first — and proposes a map out of them. Same contract: preview, then accept.
Smaller helpers on the way
- On any node: Substeps, Milestones, Metrics (KPI), Risks, Improve this node — quick AI actions in the node detail.
- On any node's task list: a proposal of concrete tasks — "The AI added N tasks — edit or delete what doesn't fit."
- In the dashboard: an AI summary of the project — a paragraph for your status report.
- In My day: a short morning encouragement (⟳ regenerates it); without AI, proverbs rotate instead.
When the AI is down
An AI outage is not a problem with your data: the buttons switch to "AI temporarily unavailable" with the explanation that the service is not responding and the rest of the app works normally — the buttons come back by themselves. Maps, tasks and sharing keep running; AI is an add-on, never a dependency.
What it costs and where it runs
- In the cloud, AI is included in the plan, with a monthly quota per instance.
- Self-hosted, you bring your own model: a local Ollama is free, or point the app at any compatible endpoint — see Configuration. Without a model configured the AI features simply stay hidden.
Next
- Automations and agents — when a machine should also execute the work, not just plan it
- MCP server — let your own AI assistant (e.g. Claude) work with your maps