Lesson 3.4: Your First AI-Assisted Task

Purpose

You've learned the theory. Now let's do it. This lesson walks you through a complete, real managerial task using AI, step by step, with judgment checkpoints at each stage.

By the end, you'll have completed your first AI-assisted task and you'll know how to replicate this process for other work.

Why This Matters for Managers

Reading about AI is one thing. Actually using it is another. This hands-on lesson bridges that gap.

You'll:

• Work through a real task (drafting a team update)
• See exactly how to apply each concept
• Practice with judgment checkpoints
• Experience the workflow end-to-end

After this, using AI becomes less intimidating and more practical.

Core Concepts

The End-to-End Workflow

Here's the complete process you'll follow:

1. Identify the task (what you actually need to do)
2. Gather inputs (information you have to give AI)
3. Write the prompt (using the framework you learned)
4. Get AI response (the first draft)
5. Evaluate (using the five-dimension checklist)
6. Refine (if needed)
7. Verify and finalize (before using it)
8. Use and reflect (learn what worked)

Practical Managerial Use Case: Complete Walkthrough

Task: Drafting a Team Update Email

Your situation:

• You manage an 8-person engineering team
• You send a weekly update email every Monday morning
• This week's update needs to cover: Q1 results (good), upcoming projects (exciting), team recognition (important to you), and a request for feedback
• Tone: Warm, direct, not corporate
• You want to feel personal, not generic

Step 1: Identify the Task

What you're doing: Drafting a team update email

What you need: First draft based on bullet points you have

Why AI helps: You have the information but composing a polished email takes 30 minutes. AI can generate a first draft in 2 minutes, then you refine it.

Realistic expectation: This will take 15-20 minutes total (including AI time and your refinement). Manual would take 30-40 minutes. Savings: 15-20 minutes.

Step 2: Gather Inputs

Before you write the prompt, gather what you need to communicate:

Q1 Results: "We shipped 8 features, hit all deadlines, shipped with zero P1 bugs. User satisfaction on new features is 4.8/5."

Upcoming projects: "We're starting the mobile app redesign—exciting but complex. Also improving the payment system for better reliability."

Team recognition: "Sarah led the feature launch with exceptional execution. Marcus mentored two junior team members through major work. Team overall showed great collaboration."

Feedback request: "I want to hear about workload, what's working well, and what could improve."

Your voice/values: "I value transparency, collaboration, and development. I'm direct but warm. I want the team to feel proud and supported."

Step 3: Write the Prompt

Using the framework (task + context + format + constraints), write your prompt:

        Task: Draft a weekly team update email
        
        Context: I manage an 8-person engineering team. We had a great Q1:
Shipped 8 features, hit all deadlines, shipped with zero P1 bugs
User satisfaction on new features: 4.8/5
Sarah led the feature launch with exceptional execution (recognition important to me)
Marcus mentored junior team members through major work
Team showed great collaboration overall
        
        I want to recognize achievements, share what's coming next (mobile app redesign and payment system improvements), and ask for feedback on workload and ways I can support them better.
        
        Format:
3-4 short paragraphs
Professional but warm tone, not corporate
Include: What we accomplished, what's coming, recognition, feedback request
        
        Constraints:
No jargon or corporate language ("synergies," "moving the needle," etc.)
Should sound like it's from me personally, not generic
Should make the team feel proud and supported, not stressed
Maybe 250-300 words
Friendly and direct, not overly formal
        

Step 4: Get AI Response

You paste your prompt into your AI tool. Here's what AI returns:

        Subject: Q1 Wins + What's Ahead
        
        Team,
        
        Great work wrapping up Q1. We shipped 8 features on schedule, maintained zero P1 bugs, and customers are loving what we built—4.8/5 on new feature satisfaction is genuinely excellent. That doesn't happen without focus and collaboration, so thanks.
        
        A couple of recognitions: Sarah, your execution on the feature launch was exceptional. You kept the team organized and shipped something customers actually want to use. And Marcus, your mentoring of our junior folks through major work made a real difference in their growth. Both are examples of the kind of work that makes teams strong.
        
        What's coming: We're starting the mobile app redesign—this is big work but exciting because we get to rethink the user experience. We're also improving the payment system to make it more reliable. Both have real impact on our users.
        
        I want to make sure I'm supporting you well. Please let me know: How's your workload? What's working well? What could I do differently? I'm reading feedback carefully and want to make sure we're building a place where you can do great work and grow.
        
        Thanks,
        [Your name]
        

Step 5: Evaluate (Using the Five-Dimension Checklist)

Read the AI output and evaluate:

1. Accuracy: All facts correct? Q1 results accurate, recognitions match what you said? Yes, all accurate.

1. Tone: Does this sound right for your team and values? It's warm, direct, and not corporate. It feels personal. Good tone.

1. Completeness: Did it include everything? Q1 wins (yes), upcoming projects (yes), recognition (yes), feedback request (yes). Complete.

1. Context: Does it reflect your specific team and situation? It does—feels like it's addressing your actual team, not generic team.

1. Values/Authenticity: Does this sound like you? Would you send this? Actually, it feels close. The tone matches how you want to communicate. You'd use this.

Evaluation summary: Good output. No major issues. Minor refinements possible, but it's close to ready.

Step 6: Refine

Where could it be better?

• The opening paragraph is good but could feel slightly warmer
• "Please let me know" in the feedback paragraph feels a bit formal for your voice. You'd probably say "I'd love to hear"
• The closing is fine, but maybe add a personal touch about looking forward to the projects

Your refinement:

You have two options:

1. Ask AI to refine (send back: "Good draft. Make it warmer and less formal. Change 'Please let me know' to something more conversational. Add excitement about the upcoming projects.")
2. Refine manually (just edit it yourself, which might be faster for small changes)

For this case, you decide to do minor manual edits:

Original: "Great work wrapping up Q1."

Your version: "Impressive Q1, everyone. We shipped 8 features on schedule, maintained zero P1 bugs, and customers are genuinely happy with what we built."

Original: "Please let me know: How's your workload?"

Your version: "I'd love to hear: How's your workload?"

Original: "Thanks for the great work" → Add: "Really excited about the mobile redesign and the payment system work. These are big bets for us, and I know they'll push us to do our best work."

Step 7: Verify and Finalize

Final check before sending:

Read the refined email one more time:

• Does it sound like you? Yes.
• Is everything accurate? Yes.
• Would you be comfortable with this sent to your team? Yes.
• Is there anything you'd be embarrassed about if someone questioned it? No.

Final decision: Send it.

Time spent:

• Gathering inputs: 5 minutes
• Writing prompt: 3 minutes
• AI response: 1 minute
• Evaluation: 2 minutes
• Refinement: 3 minutes
• Final verification: 1 minute
• Total: 15 minutes

Comparison: Writing this email from scratch would probably take 30-40 minutes. You saved 15-25 minutes and probably have a better-structured email.

Step 8: Use and Reflect

After you send it:

Within 24 hours, you reflect:

• Did it land well? Team responded positively.
• What worked? The warmth and recognition felt good to them.
• What would you do differently? Next week, you might ask AI to handle even more of the draft (less refinement needed).
• Did the time savings matter? Yes—freed up time you used for strategic planning.

Variations: Other Tasks Using This Workflow

The same end-to-end workflow applies to other tasks. Here's how it looks for different scenarios:

Variation 1: Summarizing a Meeting

Steps 1-2: You have a 60-minute meeting transcript. You need a 10-minute summary for stakeholders who weren't there.

Step 3 (Prompt): "Summarize this meeting transcript in 8-10 bullet points covering: (1) Decisions made and by whom, (2) Action items with owners and deadlines, (3) Risks identified. Focus on what matters for someone who wasn't there. Skip small talk and details."

Step 4: AI generates structured summary.

Step 5: Verify—Did it miss any key decisions? Add anything important that's missing.

Step 6-7: Minor edits, then send.

Time: 5-10 minutes (vs. 30 minutes to do manually).

Variation 2: Brainstorming Solutions

Steps 1-2: You're thinking about how to improve your team's code review process. You have some initial thoughts about what's wrong.

Step 3 (Prompt): "We want to improve our code review process. Current state: 2-3 day review turnaround, junior devs sometimes wait for senior reviews, reviews are sometimes surface-level. Constraint: We don't want to add more meetings. Generate 10 ideas for improving the process. Be creative."

Step 4: AI generates 10 ideas.

Step 5: Evaluate—Some are obvious, some are interesting. Which ones could actually work in your culture?

Step 6: Pick the 3 most interesting, refine them.

Step 7: You now have a concrete brainstorm to present to the team.

Time: 15-20 minutes (vs. 1-2 hours thinking about this alone).

Variation 3: Drafting Feedback

Steps 1-2: You're preparing feedback for an employee. You have notes on what you want to address.

Step 3 (Prompt): [Describe the situation and feedback you want to give]

Step 4: AI generates a structured approach or draft talking points.

Step 5-6: Evaluate for tone, appropriateness, completeness. Refine heavily because this is high-stakes.

Step 7: Use the refined version as your preparation, then deliver the conversation in person.

Important: For high-stakes feedback, AI drafts + heavy customization + personal delivery is the right approach.

Time: 20-30 minutes (vs. 30-60 minutes thinking through how to approach it).

Anti-Patterns / Misuse Risks

Misuse Risk 1: Expecting First Draft to Be Perfect

"AI drafted something. I'll send it as-is."

Why it fails: First drafts are rarely perfect. Refinement is expected.

Misuse Risk 2: Skipping the Evaluation Step

"I got output. Time to use it."

Why it fails: You might catch problems that damage credibility later.

Misuse Risk 3: Overcomplicating Simple Tasks

"I'll use AI even for simple tasks that take 5 minutes."

Why it fails: Learning overhead makes it slower.

Human Judgment Checkpoints

As you complete your first AI-assisted task, check:

1. Is this task appropriate for AI? (Something you identified in Chapter 2?)
2. Did I provide enough context? Could AI understand your situation?
3. Did I evaluate thoroughly? Or did I skip steps?
4. Did I customize the output? Or send generic AI?
5. Did this actually save time? Or was the learning curve too steep?

Responsible AI Considerations

Taking Accountability

You're responsible for whatever you send, even if AI helped. The verification step ensures you stand behind the output.

Learning as You Go

Your first task might take 20 minutes. Your 20th task might take 5 minutes. The time savings come with practice.

Reflecting on What Works

After each AI-assisted task, spend 2 minutes reflecting: "What worked? What didn't? What would I do differently?"

Practice / Reflection Prompts

1. Your First Task: What's a real task you could complete using this workflow in the next week?

1. Complete the Workflow: Walk through all 8 steps for your task. Document each step.

1. Timing: How long did each step take? What took longer than expected?

1. Evaluation Honesty: When you evaluated the output, what problems did you catch? How serious were they?

1. Refinement: How much time did refinement take? Was it worth it?

1. Did it Help: Did AI actually save time? Or did the learning curve offset savings?

1. Next Task: What will you do differently on your second AI-assisted task?

Key Takeaways

1. Follow the 8-step workflow. Identify → Gather → Prompt → Response → Evaluate → Refine → Verify → Reflect.
2. Build judgment checkpoints in each step. Don't skip verification.
3. Expect to refine. First drafts are drafts.
4. Time varies by task. Email: 15 min. Summary: 10 min. Complex task: 30 min.
5. You're responsible for the output. Verification is how you maintain that.
6. Learn as you go. First task is slower. Later tasks are faster.