👋

Welcome

Week 2

The First Pancake

Everyone knows the first pancake is a write-off. Too much batter, wrong heat, weird shape. You look at it, you know what happened, and you flip it anyway.

You don't throw the pan away. You don't decide you can't cook. You just make the next one a little better.

Getting a good result from AI works exactly the same way. Your first prompt probably won't land. That's not a problem — that's the process. You adjust, try again, and slowly it gets closer to what you actually wanted.

The first pancake is never the one you serve. But you need it to get to the good ones.

This week is about embracing the first pancake.

🎬

This Week's Concept Drop

🤸

Your Weekly Experiment

Pick a real work task and use AI to complete it, but start with a deliberately rough prompt. Then have another go at least three times, refining your prompt each round.

The goal isn't to produce something perfect. The goal is to feel the difference between "trying to get it right on the first try" and "treating refinement as the actual work."

Choose your track:

🧑‍💼

Hands-On Track

Your Experiment:

Pick one work task where you'd normally aim to get it right on the first try. Examples:

  • Drafting an email to a stakeholder
  • Summarizing a meeting or document
  • Brainstorming solutions to a problem
  • Analyzing a dataset or trend

Step 1: Start Rough (2 minutes max)

Give AI a deliberately incomplete or vague prompt. Don't overthink it. Examples of "rough":

  • "Write an email about the project delay" (no context, no tone, no audience detail)
  • "Summarize this report" (without saying what aspects matter)
  • "Give me ideas for solving [problem]" (without constraints or context)

Let the output be bad. Don't start over. Don't apologize to the AI. Just look at what it gave you.

Step 2: Have another go (3 rounds minimum)

Now improve it. Each round, add one thing:

  • Round 1: Add context or constraints ("Make it more direct" or "Focus on the financial implications")
  • Round 2: Challenge the output ("What's missing from this?" or "What assumption is this making?")
  • Round 3: Refine toward your actual need ("Make this sound less formal" or "Add a specific example")

Step 3: Reflect (before you talk to Eko)

Compare your rough prompt to your final prompt. What changed? What did you learn about how to ask better questions?

Success looks like: You feel the difference between "trying to nail it on the first try" and "treating the first try as raw material."
👔

Team Lead Track

Your Experiment:

Pick a work task you do regularly: a report, a roster, a toolbox talk, an email, a brief, a plan. Something real. Examples:

  • A decision you need to make
  • A process that's breaking down
  • A goal you're not hitting
  • A conflict or tension in the team

Step 1: Start Vague (2 minutes max)

Give AI an intentionally incomplete prompt. Examples:

  • "My team is struggling with [problem]" (no context about why or what you've tried)
  • "We need to decide whether to [option A] or [option B]" (no criteria, no constraints)
  • "How do we improve [metric]?" (no root cause analysis, no context)

Let the output be generic. That's the point. You're practicing starting without having it all figured out.

Step 2: Add the real details. Your team's situation, priorities, or constraints.

Each round, add something that sharpens the thinking:

  • Round 1: Add context ("Here's what we've tried so far..." or "Here's why this matters...")
  • Round 2: Add constraints or criteria ("We have [X constraint]" or "Success looks like [Y]")
  • Round 3: Ask AI to stress-test ("What are the risks with this approach?" or "What am I not considering?")

Step 3: Tell it who this is for, and what they care about.

Look at your final prompt vs. your first prompt. What did you clarify for yourself (not just for the AI) through this process?

Success looks like: You realize that having another go isn't about "fixing" the AI. It's about clarifying your own thinking.
💡 Psst, need help?

If you get stuck during refinement, try these:

🤔 Can't think of what to refine
Ask AI: "What's missing from this?" or "What questions should I answer to make this better?" Ask yourself: "What would I need to add if I were explaining this to a colleague?"
📉 Refinements aren't improving the output
You might be adding detail without adding clarity. Try asking AI to challenge the output instead of just expanding it. Example: "What's weak about this approach?" or "What assumption is this making?"
😰 Feels like wasting time
Good. That discomfort is the point. You're learning that refinement feels slow but is actually faster than trying to be perfect on the first try. Stick with it for at least three rounds.
🔁 Still generic after 3 rounds
Add more specific context about your situation, constraints, or goals. Or try: "Why is this output generic? What context am I not providing?"

⚡ Quick tips

Start rough on purpose. Refine, don't restart. Use disagreement as data. Ask meta-questions: "Why did you interpret it that way?"

🧠

Time to Reflect with Eko

After you complete your experiment, it's time to unpack what happened. This is where you move from doing to understanding.

🤖

Chat with Eko

Your AI reflection coach

💡 Pro tip: Start your reflection with this prompt:

"Eko, I just completed my Week 2 experiment for the AI Starter Sprint. I practised refinement by starting with a rough prompt and having another go at least three times. Can you help me reflect on what I learned?"

Important: Eko isn't here to tell you what to think. Eko is here to help you think. If a question doesn't land, say so. If you need to go deeper on something, ask.
🤓

The Deep Dive

Not required, for those who want to go deeper.

1

Kolb's Experiential Learning Cycle

Article

The framework behind "learn by doing." Kolb breaks learning into four stages, and most of us skip the two that matter most. This will help you understand why reflection after your experiment isn't optional.

📄 Read the Article →
2

Executive Functions & Self-Regulation

Video

Self-regulation isn't about willpower. It's a cognitive skill that determines how well you adapt when your usual approach stops working. This video explains the mechanics behind why change feels so hard, even when you know it's necessary.

▶️ Watch on YouTube →
3

ReThinking with Adam Grant

Podcast

Adam Grant literally wrote the book on rethinking. This podcast is a masterclass in questioning your own assumptions, exactly the muscle you're building this sprint. Start with any episode that grabs you.

Spotify Apple Podcasts
← Back to Week 1 Next: Week 3 →