Write One-on-One Talking Points with AI
Five minutes before a 1:1, paste what's on your mind into @vustbot — wins, friction, asks, last meeting's leftovers — and request prioritized talking points with a suggested opener for the hard item. The model organizes your own raw material into a conversation plan; nothing about the other person is fetched from anywhere.
The worst 1:1s aren't hostile — they're empty, both sides improvising pleasantries for thirty minutes. The fix is embarrassingly cheap: five minutes of dumping your actual topics into a model and letting it sequence them, hard thing included.
What the AI does in this scenario
- Raw brain-dump in, prioritized conversation plan out
- The difficult topic gets a suggested opener, not a euphemism
- Carry-over slot for what you promised last time
- Works both directions: manager preparing, or report managing up
Worked example: write one-on-one talking points with ai
01Input — what you paste
For 1:1 with my manager: shipped the migration early. annoyed that my design doc sat unreviewed 2 weeks. want to ask about leading the payments workstream. also she mentioned a reorg rumor last time, follow up.
02Output — what comes back
Talking points (30 min): 1) Open with the win — migration shipped early, one sentence, don't over-sell. 2) Follow-up owed: the reorg rumor she raised. 3) The ask: payments workstream lead — tie it to the migration result. 4) The friction (opener: "Can we talk about review turnaround? My design doc waited two weeks — what would make that faster?"). Close: agree one concrete next step on #3.
How to write one-on-one talking points with ai — step by step
- 1Brain-dump without editing
Everything on your mind about this person and this working relationship, unsorted, into @vustbot. Ordering while dumping is exactly the cognitive load you're delegating.
- 2Ask for priority plus one opener
Have the model sequence the items for a real conversation arc — win, business, the ask, the hard thing — and draft an opening sentence for whichever item you're most likely to chicken out of.
- 3Bring three points, not ten
Cut the generated list to what fits the slot; a 1:1 is a conversation, not a review meeting. The discarded items seed next time's dump.
AI vs doing it manually
Preparing 1:1s manually works when you already keep a running topics doc — many great managers do, and a model adds little to that discipline. The realistic case is the other 80%: walking to the meeting assembling thoughts in the corridor. There, the AI's contribution isn't insight — it's sequencing and nerve. It puts your buried "I want the payments lead" third instead of never, and hands you a non-confrontational opener for the friction item. The content was always yours; what the model removes is the improvisation tax.
The prompt to copy
Turn these raw notes into talking points for a [LENGTH]-minute one-on-one with my [MANAGER/REPORT]. Prioritize into max 4 items with a suggested order: quick win first, hardest item second-to-last. For the most difficult item, draft a neutral opening sentence. Add a "carry-over" slot if my notes mention past promises. My notes: [PASTE RAW THOUGHTS]
Frequently asked questions
Should I paste sensitive feedback about a colleague into the bot?
Keep it functional: "friction on review turnaround" prepares the conversation just as well as a named rant, and your talking points shouldn't read like a dossier anyway. Paste situations and behaviors, not personnel-file judgments — the model sequences topics; the sensitive framing stays in your head.
Can AI prepare talking points for a 1:1 where I have to give hard feedback?
It can structure it — situation, behavior, impact, then a question — and draft the opener, which is genuinely the hardest sentence. What it can't do is calibrate to the relationship: how direct this person can hear. Take the structure and the opener; deliver the middle in your own words.
What if my 1:1 notes are just three vague feelings?
Paste the three vague feelings — "annoyed about reviews, worried about the reorg, want more scope" is a complete input. Ask the model to turn each into a concrete discussable item with a question attached. Vague-to-specific is a transformation models do well; blank-page-to-anything is the part that was blocking you.
Related in Meetings
Try it on your real task
Send the prompt above with your own facts and judge the output yourself.
Open @vustbot