Skip to content

AI Agents for Leaders Who Have No Time to Learn AI · Section 4 · Resources

Send messages your colleagues recognize as yours, Complete Resource Pack

Every request, file and link from the lectures in one place. Copy a request with one click.

Arun Nagarathanam · 20 Oct 2026 Guide · 26 of 42 in this course

Lecture 4.1 · Leadership Communication

Can AI help while you still sound like you?

See why a fluent AI draft can still be wrong for the person whose name goes on it, and what useful writing help has to keep.

Ideas to keep

  1. 01Marcus decided on a 3-week trial: routine status written in the project folder before Tuesday, a 30-minute meeting on decisions, and one owner for each unresolved blocker.
  2. 02The fluent AI draft announced that the new format will eliminate cross-team coordination delays across the department. Marcus never decided that.
  3. 03Clearer wording helps. Changing what you are committing to puts your credibility at risk the moment the next normal project delay occurs.
  4. 04Colleagues organize their work around the commitments in your messages, and they listen for your recognizable manner of speaking to judge whether you mean it.
  5. 05Typing "make this sound professional" cannot keep your voice, because the software does not yet know what your voice sounds like.

Try it (optional)

  1. Pick one message or announcement that carried your name this week.
  2. Write down the decision and any condition behind it in one or two lines.
  3. Read the message again and check whether any sentence promises more than that decision.

Files and links

Marcus is a prepared fictional example written for this course. The inflated draft shows the kind of drift to look for; it is not output from a recorded AI run.

README.md

Lecture 4.2 · Your AI Voice Skill

Why does the draft sound nothing like you?

Teach AI how you write from three of your own messages, curate the rules it proposes, and save them as a voice skill.

Ideas to keep

  1. 01Instead of adjectives like "warm and professional", Marcus supplies 3 short permitted updates he wrote himself and a spoken explanation of what he likes and dislikes.
  2. 02Ask the agent to quote the line from your messages that supports each proposed rule, because an agent can treat an isolated phrase from one old email as a permanent habit.
  3. 03Marcus keeps the rule about opening with the decision and its reason, changes the rigid 12-word limit, and removes the one-off audit phrase.
  4. 04No colleague names or past project dates stay inside the general writing rules.
  5. 05A saved skill that captures your curated writing preferences is called a voice skill. The free starter pack names it personal-voice.

Try it (optional)

  1. Open practice/leadership-communication.html and go to the 01 Voice skill curation tab.
  2. Read the three proposed rules with their quoted evidence and Marcus's Keep, Change and Remove decisions.
  3. Find the sample that the removed rule came from in practice/marcus-communication-files.md.
  4. If you want to try it, gather 3 short messages you are permitted to share and run the first request below in an AI product your employer permits.
Request · Propose voice rules with evidence
Here are 3 permitted updates I wrote to my team, along with my spoken explanation: I like to state the decision and the reason for it in the first paragraph, I prefer everyday words over corporate jargon, and I always make clear what action I need from the reader.

Propose a short list of recurring writing rules from these examples and my explanation. Next to each proposed rule, quote the wording from my examples that supports it so I can decide what to keep, change, or remove.
Use in: ChatGPT or another AI product your employer permits, with your 3 messages attached
Request · Curate the rules and save personal-voice
Keep the first rule about opening with the confirmed decision and the practical reason behind it. Change the second rule so it prefers clear, concise sentences without imposing a rigid 12-word cap that removes the reason behind a request or a condition on a decision. Remove the phrase "effective immediately until audit closure" and any past project names or dates so no one-off incident details enter the reusable skill. Apply these curated rules to my weekly-meeting draft, and save the accepted guidance as my personal-voice skill.
Use in: The same conversation, after reading the proposed rules

Files and links

The three proposed rules are a prepared illustration. Use only messages you are permitted to share with your AI product.

practice/leadership-communication.htmlpractice/marcus-communication-files.mdprompts/section-04-prompts.mdskills/personal-voice/SKILL.md

Lecture 4.3 · AI Drafting

How do you explain the same decision to different people?

Turn one confirmed decision into a draft for your team and a draft for a senior sponsor, without changing what you promised.

Ideas to keep

  1. 01Marcus's decision record has 4 facts: status in the project folder before 9 a.m. on Tuesdays, a 30-minute meeting on decisions and blockers, one named follow-up owner per blocker, and a review after 3 weeks.
  2. 02The same decision creates different work for each reader. The team needs when, where and how; the sponsor needs the reason, the accountability and what will be measured.
  3. 03Vary the emphasis and level of detail for each reader. Keep the decision, its conditions and its commitments the same in both messages.
  4. 04The sponsor draft first said the format "will keep delivery on schedule". Marcus changes it, because that edges toward a promise he never made.
  5. 05The team draft keeps early warnings welcome: "If you are stuck on something, bring it on Tuesday, even if you do not have a fix yet."

Try it (optional)

  1. Open the 02 Team & sponsor drafts tab in practice/leadership-communication.html.
  2. In the sponsor draft, check that the only promise is the 3-week check on whether the meeting still runs over.
  3. In the team draft, find the sentence that keeps raising an unresolved blocker early welcome.
  4. Before approving drafts of your own, check that each reader can tell what changed, what response you need and what you have not decided yet.
Request · Draft for the team and the sponsor
Use my confirmed decision record and my personal-voice skill to prepare 2 announcements for changing our Tuesday planning meeting:
1. A message to my 8 team members explaining when the change starts, how to post routine status updates in the project folder before 9 a.m. on Tuesdays, and how to bring unresolved blockers to the 30-minute meeting.
2. A message to my senior sponsor explaining the operational reason for the change, how blocker accountability is protected with a single named owner, and what we will check after the 3-week trial before making the format permanent.
Keep the underlying decision, conditions, and commitments consistent across both drafts.
Use in: ChatGPT or another AI product your employer permits, with your decision record attached
Request · Check both drafts against the decision
Compare both drafts against my confirmed decision record. Check that the sponsor draft promises only the 3-week check on whether meeting spillover dropped, with no guarantee about project milestones. Check that the team draft makes clear that raising an unresolved blocker early is welcome, even when the person does not yet have the fix. Tell me if either draft changes a commitment.
Use in: The same conversation, after reading both drafts

Files and links

Both drafts are prepared teaching examples and stay unsent.

skills/leadership-communication/SKILL.md

Lecture 4.4 · AI Rehearsal

Can you explain your recommendation under pressure?

Turn a written recommendation into a short spoken opening, then answer plausible questions aloud to find the gap before the real meeting.

Ideas to keep

  1. 01Reading your slide notes word for word rarely answers a direct question about why you rejected another option.
  2. 02Marcus asks for a short spoken opening built from his decision record, then 3 plausible questions the sponsor might ask: why now, what the team gains and how he will know the trial worked.
  3. 03The questions are plausible challenges drawn from the tension in your proposal, not a prediction of what a specific person will say.
  4. 04Answer aloud first, then read the agent's suggestion. You can turn on the microphone or type.
  5. 05Marcus's first answer was about his team's frustration, not measurement. He knows the meeting runs over but has never counted by how much, so he plans to check the last 4 Tuesday meetings.

Try it (optional)

  1. Open the 03 Spoken rehearsal tab in practice/leadership-communication.html.
  2. Read the challenge question, then the first spoken attempt and what it left out.
  3. Read the revised spoken answer and find the step that gives the 3-week check something to compare against.
  4. If you have a recommendation coming up, run the first request below and answer one question aloud before reading the suggestion.
Request · Spoken opening and plausible questions
Using my decision record for the 3-week Tuesday meeting trial, help me prepare a concise 60-second spoken opening that states the change, the reason for it and the 3-week check. Then ask me 3 plausible questions the senior sponsor could raise in the weekly delivery review, such as why now, what the team gains and how I will know it worked, so I can practice answering them aloud before seeing your suggested response.
Use in: ChatGPT or another AI product your employer permits, with your decision record attached
Request · Revise the answer after the gap
My spoken answer explained why my team dislikes the long status round but not how I will measure the change. Revise the spoken explanation to say that the meeting has been running past its 30 minutes and the new format is meant to bring it back inside 30 minutes. Flag that I do not yet have a starting number, and add a step to check how long each of the last 4 Tuesday meetings actually ran before the review.
Use in: The same conversation, after your spoken answer

Files and links

The questions and answers are a prepared rehearsal, not a recording of a live exchange.

skills/conversation-rehearsal/SKILL.md

Lecture 4.5 · Leadership Communication

Would you put your name on this message?

Apply one signature test to any AI-prepared message, and see where a one-way message stops being enough.

Ideas to keep

  1. 01The test: does this draft preserve my actual commitment, reflect my voice, and tell this reader what they need to know without hiding the tradeoff?
  2. 02Marcus's drafts kept the 3-week trial condition, made each reader's requested action clear and sounded like his direct, considerate voice.
  3. 03In his rehearsal, speaking aloud replaced a story about frustration with a plan to measure the change.
  4. 04When a senior analyst replies that 9 a.m. clashes with her early client call, another polished paragraph repeating the rule misses what the moment requires.
  5. 05As a leader, you own both the meaning of the message you send and the space for the other person's response.

Try it (optional)

  1. Take one AI-prepared message you are about to send.
  2. Check it against the three parts of the signature test.
  3. Ask yourself whether the situation needs a reply from the other person before anything is settled.

Files and links

Two-way people conversations, such as addressing performance, coaching a skill gap or assigning ownership, are the subject of Section 5.

skills/communication-and-people-leadership/SKILL.md

More in this section

Keep going

Keep building

More from Aruntastic

Practical courses for people who want to build useful things with AI.

Toolkits and programme

Go deeper

Claude certification practice

Preparing for a Claude certification?

Practice exams written by Arun: 600 questions across 4 mock exams and 2 drill banks. Unofficial preparation, not affiliated with Anthropic.

Keep going

This resource is part of a six-section course.

Follow the lectures on Udemy, then come back here whenever you want the files, requests and notes for a section.

Open the course AI Agents for Leaders Who Have No Time to Learn AI · Udemy