Thoughts
A thought is Lucille's word for a single small claim about you or your week, backed by evidence. Thoughts are how Lucille remembers things between conversations.
What a thought looks like
Each thought is a short factual sentence plus context. Examples:
- "You worked 32 hours this month." (evidence: Clockify data)
- "Your meditation streak broke yesterday." (evidence: habit tracker)
- "Most of your work time went to Project X." (evidence: Clockify rollup)
A thought is not a task, a notification, or a plan. It's a fact Lucille is keeping in mind. Tasks and plans (the things you act on) get built out of thoughts when you ask for a day brief or a plan.
Where you see them
- In the day brief: when you ask "plan my day" or get the morning summary, Lucille surfaces a small number of recent thoughts. See Day brief.
- When you ask: "what do you know about my week" / "what should I know" ā Lucille pulls relevant thoughts and replies with them.
- As reactions: when Lucille is retrieving a thought to answer you, you'll see the š¤ reaction on your message (see Emoji conventions).
How they're created
Mostly automatically:
- From connected integrations (Clockify time tracking, calendar, habits)
- From notes you send Lucille in chat
- From check-in answers (see Check-ins)
The pipeline is opaque by design ā Lucille decides which observations are worth keeping based on the evidence she has.
Turning them off
If you want fewer thoughts surfaced in chat or in the day brief, ask Lucille directly ("don't include thoughts in my morning brief") or adjust the relevant integration's settings in the web app.
You can also disable the connected integration that produces the thought ā e.g. disconnect Clockify and you'll stop getting thoughts about your work hours.
Lifecycle
Thoughts are time-scoped. A thought about "this week" is replaced by a new one when next week's data comes in. Old thoughts don't accumulate forever ā Lucille recomputes them as fresh signal arrives.