Intermediate
AI for Team Leads
The 1:1, the update, the hard email.
Anyone can use this page. Copy the prompt and try it today.
$7.99
Intermediate
The 1:1, the update, the hard email.
Anyone can use this page. Copy the prompt and try it today.
$7.99
Everyday AI
The 1:1, the update, the hard email.
You stay the editor. The model drafts. You decide.
$7.99 · 20 pages
Intermediate shelf · people notes · you still coach
CONTENTS
Read pages 3 through 6. Then copy one cream box.
THE IDEA
A chat can tidy notes, shape an update, or shorten a difficult email. It was not in the 1:1. It did not see the missed handoff.
It was not in the room. If a draft adds a motive, judgment, promise, or fact you did not supply, delete it. Smooth prose does not make a guess true.
A prompt is a bounded job. Role + Task + Details + Format. Give it facts you already own, the audience, the limit, and the shape.
The useful jobs are small. A five-line 1:1 plan. Praise from one observed moment. A decision recap. A clearer hard email.
You remain the lead. The model does not rate people, diagnose conflict, or run an HR process. It drafts. You check and send—or do not send.
Short version: use it for wording, not judgment.
WHAT THE LABS FOUND
These findings support a writing workflow. They do not prove that a model can manage people.
1 · OpenAI Writing is about 40% of work-related ChatGPT messages. About two-thirds of those writing messages modify text the user supplied.
2 · Anthropic Put long notes before the question. In its tests, placing the query last improved quality by up to 30% on complex long-document tasks.
3 · Anthropic Fluency Index People who iterate also show more clarification and critical-evaluation behaviors. This is correlational, not causal.
4 · Google Useful prompt parts include objective, persona, constraints, context, and response format—compressed here into Role / Task / Details / Format.
Paste facts first. Ask for a narrow edit. Inspect the answer before it reaches another person.
BEFORE YOU PASTE
Nora’s note says, “Late again. Maybe disengaged.” The late delivery is observable. “Disengaged” is a guess.
Observed: What you saw, heard, received, or agreed.
Impact: What changed, using facts you can support.
Decision: What has already been decided, and by whom.
Missing: What you still need to ask a person.
Use first names or roles. Do not paste personnel files, private reviews, medical information, passwords, compensation sheets, protected complaints, or confidential material your workplace has not approved for an external tool.
Do not ask the model to score a person, infer intent, recommend discipline, make a hiring decision, or provide HR legal advice. Use your approved human process for sensitive employment action.
Before every paste: Which words are facts, and which are my story about the facts?
THE 1:1
Ravi has scraps from last week with Mei. The launch slipped one day. Mei asked for clearer owners. Ravi promised to unblock design.
COPY THIS
Role: You are a 1:1 notes editor. Organize my notes; do not evaluate the person.
Task: Turn my existing notes into a short plan for the next 1:1.
Details: Notes: [facts, questions, promises]. Person: [first name]. Time: [25 minutes]. Show my unkept promises clearly.
Format: Five lines: follow-up; their topic; my topic; one question; one promise or next step already in my notes.
Hard nos: No score, diagnosis, review language, invented concern, or new fact. Leave a blank when notes are thin.
Then do this. Circle the promise you own. If it sounds like a review, reply: “Room notes only. No judgment. Five lines.”
Watch for: tidy notes turning into a verdict.
THE TEAM UPDATE
Keisha’s facts: the search fix shipped, two interviews happened, and the handoff document is still open. She needs an update, not a victory speech.
COPY THIS
Role: You are a plain-language team-update editor. I remain the sender.
Task: Turn my real list into a concise weekly update.
Details: Audience: [team and manager]. Completed: [facts]. Open: [facts]. Needed: [named asks]. Next week: [confirmed items].
Format: Headings: Done, Open, Need. Short bullets. End with one factual next-week sentence.
Hard nos: Do not invent progress, metrics, risks, owners, deadlines, or wins. Do not call unfinished work momentum.
Then do this. Mark every bullet you can prove. Delete the rest. If it spins, reply: “Only completed, open, and needed.”
Watch for: uncertain work sounding settled.
THE HARD EMAIL
Tomas must say the date moved to Thursday. His rough draft hides the change under six lines of throat-clearing.
COPY THIS
Role: You are a hard-email editor. Keep my plain voice.
Task: Draft a short email from my facts and ask.
Details: Audience: [group]. Change: [Thursday]. Context: [design in review]. What stays: [facts]. Ask: [hold Friday; attend Thursday at 10].
Format: Subject plus 6–8 sentences. Put the change first and the action near the end.
Hard nos: No invented cause, blame, reassurance, new date, policy, or HR language. Ask once if a required fact is missing.
Then do this. Read the first and last sentences only. If they do not carry the change and action, reply: “Lead with the date. End with the action.”
Watch for: corporate fog hiding the needed fact.
SPECIFIC PRAISE
Mei watched Andre unblock a launch at 4 p.m. when a file would not export. She wants thanks, not “rock star” language.
COPY THIS
Role: You are a praise-note editor. Write from one observed moment, not a review.
Task: Draft specific praise in my voice.
Details: Person: [Andre]. Observed: [unblocked launch when file failed]. Supported impact: [file delivered]. Channel: [private/team].
Format: Four sentences: what I saw; supported impact; what I appreciated; thank you.
Hard nos: No score, ranking, personality label, extra deed, invented quote, promotion hint, or reward promise.
Then do this. Check that the impact is known, not inferred. If generic, reply: “Use the observed moment. Remove adjectives about the person.”
Watch for: praise becoming a career promise.
CONFLICT PREP
Sam and Jules disagree about ownership of the search document. Pat needs a fair opening, not an AI verdict.
COPY THIS
Role: You are a conflict-meeting prep editor. Organize known facts and questions. Do not referee.
Task: Make a one-page prep sheet for a 30-minute conversation.
Details: People: [names]. Observed facts: [facts]. Stated positions: [label each statement]. Needed outcome: [one owner and written next step].
Format: Shared facts; disputed points; neutral opening; one question each; decision needed; follow-up.
Hard nos: Do not choose who is right, infer motives, diagnose behavior, invent quotes, assign blame, or recommend discipline.
Then do this. Cross out anything claiming what someone “really wants.” If it chooses a side, reply: “Separate facts from disputed statements. No verdict.”
Watch for: one account becoming the official story.
THE DECISION RECAP
Sofia decided in a hallway: ship Thursday and leave the new color for later. Two people missed it.
COPY THIS
Role: You are a decision-recap editor. Record a decision already made.
Task: Draft a brief recap for people who were absent.
Details: Decision: [ship Thursday]. Out: [new color]. Decision-maker: [name accurately]. Owner: [name or blank]. Timing: [confirmed].
Format: Label “Decision recap.” Five sentences: decision; what is out; owner if named; timing; where questions go.
Hard nos: No invented vote, consensus, reason, owner, option, or evidence. Do not turn it into a proposal.
Then do this. If risk is material, ask someone present to check the facts. If alternatives appear, reply: “Record the decision only.”
Watch for: invented consensus.
THE UPWARD UPDATE
Imani has two people at capacity, one open role, and a launch risk. She needs a scope decision by Wednesday.
COPY THIS
Role: You are an upward-update editor. Use only my facts and stated request.
Task: Draft a concise note about delivery, capacity, and the decision I need.
Details: Facts: [paste]. Supported risk: [what may happen and why]. Decision: [scope cut or hold date]. Needed by: [Wednesday].
Format: Current state; Capacity; Risk; Decision needed. Maximum 180 words.
Hard nos: Do not invent headcount needs, workload percentages, employee weaknesses, urgency, costs, or consequences.
Then do this. Highlight the request. It should name a real decision. If it lobbies with fiction, reply: “Facts, supported risk, one decision.”
Watch for: blaming a person to strengthen the case.
THE MEETING THAT SHOULD BE AN EMAIL
Andre’s 45-minute standup is really a status note. Nothing on the list requires a live room.
COPY THIS
Role: You are a meeting-to-email editor. Help replace a meeting when no live discussion is needed.
Task: Turn my status list into a six-line email.
Details: Purpose: [topics]. Known facts: [facts]. Open item: [item]. Owner: [name or blank]. Live need: [none/specific].
Format: Subject plus six lines: update; open item; owner; date; exception requiring discussion; cancellation line.
Hard nos: Do not invent an agenda, decision, owner, attendee, concern, or replacement call.
Then do this. If no discussion or decision is needed, send the note and cancel the invite. If it preserves the meeting, reply: “No live need is listed.”
Watch for: a manufactured reason to meet.
FEEDBACK PREP
Elena expected a brief Monday. It arrived Wednesday without the agreed risk section. “Careless” is in her rough notes, but it is not a fact.
COPY THIS
Role: You are a feedback-conversation prep editor. Describe observed work and expectations without judging character.
Task: Draft an opening and questions for the conversation.
Details: Expectation: [Monday with risk section]. Result: [Wednesday; section missing]. Impact: [review moved one day]. Next need: [owner and check-in].
Format: Three-sentence opening; two questions; next expectation; one follow-up note.
Hard nos: No personality label, diagnosis, motive, score, threat, policy claim, or disciplinary recommendation.
Then do this. Replace “you are” with “the work” or “the expectation.” If it mind-reads, reply: “Observed work only.”
Watch for: work feedback becoming identity judgment.
I WAS WRONG
Claire sent Thursday. The correct date is Wednesday. She fixed the calendar and needs a clean correction.
COPY THIS
Role: You are an apology-note editor. Keep my voice plain and accountable.
Task: Draft a short correction for my error.
Details: What I said: [Thursday]. Truth: [Wednesday]. Audience: [group]. Fix: [calendar updated]. Action: [check invite].
Format: Four sentences: I was wrong; correct fact; sorry; fix or next action.
Hard nos: Do not shift blame, invent an excuse, overstate harm, request reassurance, quote policy, or add a values speech.
Then do this. Send the correction where the wrong information appeared. If ceremonial, reply: “Own it, correct it, fix it.”
Watch for: asking the team to comfort the lead.
THE HANDOFF NOTE
Darius is away for three days. Priya needs launch facts, not his stray comment about a teammate’s mood.
COPY THIS
Role: You are a temporary-handoff editor. Keep the note operational and minimal.
Task: Turn my coverage notes into a clear handoff.
Details: Dates: [Tuesday–Thursday]. Work: [facts]. Owners: [names]. Deadlines: [confirmed]. Decisions allowed: [list]. Wait for me: [list].
Format: During coverage; Owners; Deadlines; Decisions allowed; Wait for return.
Hard nos: No private feedback, health information, mood guesses, performance judgments, invented authority, hidden task, or new deadline.
Then do this. Ask, “Does Priya need this to cover the work?” Delete anything else. If gossip appears, reply: “Operational facts only.”
Watch for: private impressions spreading through a handoff.
STUCK
The first answer may sound polished, certain, and slightly false. Use one short correction. Do not argue with it.
Invented judgment: “Use observed facts only. Remove scores, labels, motives, and conclusions.”
Added facts: “Mark every statement not present in my notes. Delete it.”
Corporate voice: “Use short sentences and ordinary words. Keep the change and ask.”
Hidden uncertainty: “Label confirmed, reported, and unknown.”
Reopened decision: “This is a recap, not a proposal. Remove alternatives.”
Invented promise: “Remove commitments I did not supply.”
HR drift: “Prepare the conversation only. Do not recommend consequences.”
Then do this. Compare the repair with your original notes. If it still fails, start a new chat with cleaner facts. Use your approved human process for sensitive employment action.
WEEK PLAN
Choose one low-risk writing job that already has clear facts. Do not paste every people problem into a model.
Monday: Make a five-line 1:1 plan. Do the listening yourself.
Tuesday: Separate done from open in the team update.
Wednesday: Send specific praise from one observed contribution.
Thursday: Put the decision or change first in a recap or hard email.
Friday: State facts, supported risk, and one needed decision in an upward update.
Keep the prompt that helped. Add one note about what it got wrong. That note is more useful than fifty templates.
Hard no: Skip AI when the real work is listening, deciding, investigating, protecting confidentiality, or taking formal action.
Then do this. Pick the message already on today’s list. Run one box, check it, and close the chat.
BEFORE YOU SEND
Jordan has a clean-looking note ready. She gives it one final pass before it reaches her team.
Fact: Can I point to the source behind every factual sentence?
People: Did it add a motive, trait, score, diagnosis, or ranking?
Promise: Did it create a commitment about timing, resources, promotion, pay, or consequences?
Audience: Is every detail necessary for this reader?
Voice: Would I say this to the person in the room?
Action: Is the next step clear, owned, and actually agreed?
Hard no: Do not send a line you cannot defend as true, necessary, and yours.
Then do this. If one check fails, edit or do not send. Keep coaching, judgment, sensitive decisions, and trust with people.
SOURCES
Chatterji et al. / OpenAI, “How People Use ChatGPT,” September 15, 2025.
Writing and editing-use figures.
https://openai.com/index/how-people-are-using-chatgpt/
Anthropic Docs, “Be Clear and Direct.”
Role guidance, long material first, query last, and the up-to-30% long-document result.
https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/be-clear-and-direct
Anthropic Academy, “The AI Fluency Index,” 2026.
Correlational findings on iteration, clarification, and evaluation.
https://academy.claude.com/tutorials/the-ai-fluency-index
Google Cloud, “Prompt Design Strategies.”
Objective, persona, constraints, context, and response format.
https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/prompts/prompt-design-strategies
No source here shows that a model can judge employee performance or replace a lead’s people decisions.