1:1 Notes, Coaching and Performance Reviews

The most sensitive AI use case for any manager is people work. Done right, AI saves you hours on 1:1 prep, coaching notes, and performance reviews — and makes you a more consistent, thoughtful manager. Done wrong, it leaks private information, produces hollow feedback, and damages trust.

This lesson gives you the safe patterns, the red lines, and four copy-paste prompts you can use this week.

What You'll Learn

• The red lines: what to never feed an AI tool about your reports
• A safe pattern for 1:1 prep and recap notes
• A safe pattern for coaching conversation preparation
• A safe pattern for performance review drafting (manager keeps the pen)
• How to use AI for pattern-spotting across a quarter of 1:1 notes
• A guide to what to disclose to your team about your AI use

The Red Lines

Before any pattern, the rules. People work involves the most sensitive data your team handles. Apply these red lines without exception.

Never paste into a general public AI tool:

• A report's full name in combination with sensitive information (compensation, health, performance concerns, conflict with a peer)
• Verbatim 1:1 transcripts containing personal disclosures
• HR-relevant information (PIP draft language, termination reasoning, harassment investigation details)
• Anything you would not want your report or HR to see in a data breach

Where it is okay to paste sensitive information:

• A licensed enterprise tool (ChatGPT Business / Enterprise, Claude Team / Enterprise, Microsoft Copilot, Gemini for Workspace) with a data-processing agreement and "data not used for training" guarantee
• A self-hosted or VPC-deployed model your company has approved
• Specialized HR-grade tools (Lattice, 15Five, Culture Amp) which have HR-specific privacy posture

Safer pattern in all cases — use initials or pseudonyms: "R" instead of "Renee," "PE3" instead of "the senior engineer on Renee's team." This is a habit worth building even in approved tools.

Always avoid:

• Letting AI write the final draft of a performance review you have not deeply edited
• Letting AI invent specific quotes or events you do not remember happening
• Using AI to "find evidence" to support a decision you have already made (this is the most dangerous use of AI in people work)

With those rules in place, the rest is upside.

Pattern 1: 1:1 Prep Notes

Before each 1:1, dump your raw thinking and let AI structure it into a concise agenda. Use initials or roles, not full names, and skip sensitive details that don't need to be in the prep doc.

You are an experienced manager preparing for a 30-minute 1:1 with a direct report. Using my raw thinking notes below, produce a focused agenda with this structure:

1. Their wins to acknowledge — 2-3 bullets, specific, grounded in last two weeks.
2. Their main project status — one sentence per active project, with my best read on where they need help.
3. Questions I want to ask — 3-5 open questions, drawing from my notes.
4. Topics I want to raise — 1-3 bullets, each phrased as a starting line (not a closed statement).
5. What I want them to leave with — one sentence.

Constraints:

• Use first-letter initials only
• Total length under 250 words
• No corporate clichés ("touch base," "circle back," "synergy")
• Frame coaching topics as questions, not directives

My raw thinking notes: [paste your bullets here]

A typical raw input is 6-8 fragmented bullets. The output is a focused 1:1 plan you can run from your phone.

Pattern 2: 1:1 Recap Notes

After the 1:1, dump your shorthand and turn it into a clean recap for your records. Keep these in a secure manager doc per report — these are sensitive over time.

You are a manager writing a private recap of a 1:1 conversation. Using my shorthand notes below, produce a structured recap with this format:

• Date: [I'll fill in]
• Themes — 2-4 bullets capturing the main topics we discussed
• Their wins / good news — bullets
• Their concerns / blockers — bullets, with what I committed to do
• Coaching opportunities I noticed — bullets, for my own follow-up only
• Action items for me before next 1:1 — bullets with due dates
• Action items for them before next 1:1 — bullets with due dates
• Watchlist — anything to revisit in 2-4 weeks

Constraints:

• Use initials, not full names
• Preserve any direct quote in quotation marks; do not invent quotes
• Flag any item I noted as confidential with [private]

Shorthand notes: [paste]

These recaps compound. After two quarters of recaps, you have an honest record of conversations, commitments, and patterns — and you stop walking into 1:1s having forgotten what you discussed two weeks ago.

Pattern 3: Coaching Conversation Prep

Coaching a struggling report is one of the highest-stakes manager conversations. AI is useful in two ways: helping you organize the conversation and helping you anticipate responses.

You are a coaching expert helping a manager prepare for a difficult coaching conversation with a direct report. The manager wants the conversation to be honest, specific, and forward-looking — not defensive or vague.

Using my notes below, help me plan the conversation:

1. The single most important message — one sentence. What is the one thing they should walk away knowing?
2. The opening (2-3 sentences) — how I should start, framed as care plus directness.
3. Specific examples to ground the conversation — bullets, from my notes. (Do not invent any examples not in my notes.)
4. The change I am asking for — concrete, observable behaviors, with examples of what "good" looks like.
5. Three likely responses from them and how I might respond — defensive, agreement, surprise.
6. The close — how I'll end the conversation and what the follow-up looks like.

Constraints:

• Use initials
• Do not soften my honest read of the situation
• Do not invent examples I did not provide
• Avoid HR jargon ("partner with," "align on," "circle back")

My notes: [paste]

Run this prep, then do the conversation yourself. AI is a thinking partner, not a script reader.

Pattern 4: Performance Review Draft (Manager Keeps the Pen)

The most dangerous mistake managers make in 2026: letting AI write a performance review from a thin prompt. The output looks polished, but it is generic, it can miss critical nuance, and it can subtly misrepresent the report.

The safe pattern: you write the substance in raw form, AI structures it into the company's required template, you review every single sentence.

You are an experienced manager turning my raw, honest notes into a structured performance review using my company's template (below). Constraints:

• Use only the content I provide. Do not add achievements, behaviors, or examples I did not include.
• Where my notes are thin in a section, write "[needs more detail from manager]" — do not fill the gap.
• Preserve any direct quote from feedback collected (peer, customer, skip-level) in quotation marks.
• Use initials for anyone named.
• Tone: direct, specific, fair, written in active voice.
• Avoid vague praise ("great team player") and vague criticism ("could be more strategic"). Both need a behavior + an example.
• Avoid loaded words flagged by performance-review research as gender-biased ("nurturing," "abrasive," "bossy," "emotional," etc.)

Template structure: [paste your company's review template here]

My raw notes on this person's work this period: [paste]

Peer / cross-functional feedback I have collected: [paste]

Then do the work AI cannot:

• Read every sentence and ask "would I say this to their face?"
• Confirm every example is real and remembered, not invented
• Check the document for hedging, padding, or vague language
• Calibrate against your other reports — does this rating mean the same thing for everyone?

A safe performance review takes 30-45 minutes with this pattern. Without AI it takes 90+. The savings are real and the quality is higher — if you keep the pen.

Pattern 5: Pattern-Spotting Across a Quarter

Once you have a quarter of 1:1 recaps and feedback notes, AI can help you spot themes you would otherwise miss.

You are a coaching analyst helping a manager identify patterns in their direct report's quarter. Read the recaps and notes below and produce:

1. Three recurring strengths — with at least two examples each from the notes
2. Three recurring growth areas — with at least two examples each
3. Inflection points — moments where their work or attitude visibly shifted
4. Stuck patterns — things I keep raising that have not changed
5. What I might be missing — areas I rarely write about that could be worth checking on

Use initials only. Quote directly from my notes when illustrating. If a section has insufficient evidence, say so — do not fabricate.

Quarterly notes: [paste]

This is the moment AI earns its keep in people work. A pattern you cannot see in 12 weeks of notes appears in 90 seconds.

What to Disclose to Your Team

Be transparent with your team about your AI use in people work. A short message to your reports:

"I want to be open about how I use AI in our work together. I use AI to help structure my 1:1 notes, organize my coaching prep, and draft sections of performance reviews. I never paste sensitive information about you into public AI tools — I use [company-approved tool] with our enterprise data agreement. The substance — what I think of your work, what we discuss in our 1:1s, your performance rating — is mine and yours, not the AI's. If you ever want to know how I used AI for something specific to you, ask."

That message removes the awkwardness, sets the trust line, and invites them to ask questions.

Key Takeaways

• People work is the most sensitive AI use case for managers — apply the red lines without exception
• Use initials or pseudonyms by default, even in approved tools
• Five safe patterns: 1:1 prep, 1:1 recap, coaching prep, performance review draft, quarterly pattern-spotting
• The manager keeps the pen on performance reviews — AI structures, you write the substance
• Never use AI to "find evidence" supporting a decision you have already made
• Disclose your AI use to your team openly; transparency is the trust line