Guides / Guide

Prompt templates that turn scattered notes into a 1:1 you're actually ready for

By Michael Shreeves · 13 June 2026 · Leadership & Management · 5 min read

Background

Most 1:1 prep happens in the two minutes before the call, scrolling back through a chat thread trying to remember what was open last time. That's not a preparation problem you fix by trying harder - it's a format problem. The information usually already exists: a scrap of notes from last week, a few Teams messages, maybe a ticket or two the person's been working. It's just scattered across three places and nobody's pulled it into one shape before the call starts.

Handing an AI tool that same scattered mess and asking for "a 1:1 agenda" gets you a generic list - "discuss progress, discuss blockers, discuss wellbeing" - because that's the safest thing to generate when it hasn't been given anything specific to work with. The fix is the same one that works everywhere else on this site: give it the actual raw material and ask for a specific structure, not a topic to riff on.

Setup

The shape every template below follows:

  • Role - a manager prepping for a real conversation, not writing a performance review.
  • Context - what's being handed over (last time's notes, anything since) and what should come back (an agenda you can glance at during the call, not a report to read beforehand).
  • Inputs - the actual scraps: last 1:1's notes, anything logged since, open items you already know about.
  • Task - carry forward anything unresolved, surface what's changed, and flag anything that's gone quiet that shouldn't have.
  • Constraints - short enough to glance at mid-call, no invented topics, nothing that reads like it was generated for a stranger.

Standard weekly 1:1

For the ordinary case - a recurring 1:1 with someone you manage day to day:

You're helping me prepare for a regular 1:1 with someone I manage.

Context: I'll give you my notes from the last 1:1 and anything relevant
since. Return a short agenda I can glance at during the call - not a
report, a working list.

Inputs:
- Last 1:1's notes: [PASTE OR SUMMARISE]
- What's happened since (messages, tickets, updates you're aware of):
  [LIST]
- Anything you already know you want to raise: [LIST, OR "NOTHING SPECIFIC"]

Task: carry forward anything open from last time that hasn't clearly been
resolved. Add anything that's changed since and is worth raising. Flag,
separately, anything that's gone quiet since last time that you'd expect
to have moved - that's often the thing worth actually asking about.

Constraints: five items or fewer. Plain language, no invented topics I
didn't give you. Nothing that reads like corporate templating - this is
for one specific person, not a generic direct report.

Skip-level or infrequent 1:1

For someone you don't see every week, where "what's changed since last time" spans a longer gap and you need more context to ask a good question rather than a generic one:

You're helping me prepare for a 1:1 with someone I don't manage day to
day and haven't spoken with one-to-one in a while.

Context: return three or four specific, well-informed questions - not a
generic "how are things going", but questions that show I know what
they've actually been working on.

Inputs:
- What I know they've been working on: [LIST, EVEN IF PATCHY]
- Anything raised about them by their manager or in team updates: [LIST]
- The last thing I remember discussing with them directly: [DESCRIBE, OR
  "CAN'T REMEMBER"]

Task: turn what I know into specific questions, not generic ones. If I've
given you very little, say so plainly rather than inventing detail to fill
the gap - a short honest list beats a long fabricated one.

Constraints: no invented facts about their work. If something I've given
you is ambiguous, ask me to clarify rather than guessing at what I meant.

Turning scattered feedback into a team update

For summarising a spread of feedback - survey comments, retro notes, informal remarks picked up over a few weeks - into something you can actually present back to a team without it reading as either whitewash or a pile-on:

You're turning a spread of team feedback into a short, fair update to give
back to the team.

Context: return three to five themes, each with a plain description and,
where there's a clear one, a next step - not a list of individual quotes
attributed to individuals, and not a single averaged-out non-statement.

Inputs:
- Raw feedback, however it came in (survey text, retro notes, remarks):
  [PASTE]
- Anything already being actioned that this should acknowledge: [LIST, OR
  "NOTHING YET"]

Task: group the raw feedback into genuine themes rather than restating
each comment. Where feedback is mixed or contradictory on the same topic,
say so rather than picking a side or averaging it away. Don't attribute
any comment to an individual, even if the source text names one.

Constraints: no invented next steps I haven't actually committed to. If a
theme has no realistic next step yet, say that plainly rather than
manufacturing one to sound complete.

Edge Cases & BCP

  • A model asked to "summarise feedback" will smooth over genuine disagreement into a single bland theme. If two people said opposite things about the same issue, that's worth knowing - ask explicitly for contradictions to be surfaced, not resolved.
  • Never paste identifiable feedback with names still attached into a general-purpose AI tool unless you've checked your organisation's actual policy on it - strip names and identifying detail first, the same rule that applies to any HR-adjacent text.
  • A generated agenda that looks complete can hide a gap you didn't feed it. If you only gave it last week's notes and nothing about what's happened since, it has no way to know something important changed - the agenda is only as current as what you actually typed in.
  • Watch for it inventing a next step that sounds plausible but was never agreed. A commitment that was never actually made, written down as if it was, is worse than no agenda at all - check every action item against what was genuinely decided before you rely on it in the room.
  • This is a memory aid, not a record. Keep whatever your organisation actually uses for 1:1 notes as the source of truth - the generated agenda is disposable prep, not the thing you'd point to later if someone asked what was discussed.

← All guides