Session 3: ChatLLM Power Moves -- Beyond Basic Questions

Facilitator Guide: Session 3

Duration: 75-90 minutes. By now they have used ChatLLM a few times. This session takes them from "I can use it" to "I can use it well." Expect more confident questions. Push them further.

Opening: Homework Review (15 minutes)

Facilitator: This review is the most important part. Their real-world experience is the richest learning material you have.

Ask: "Tell me about the three times you used ChatLLM this week. What worked? What did you struggle with? What surprised you?"

Listen for patterns. Common themes at this stage:

• "The first answer was not great, but when I gave it more detail, it got way better." (They are learning iteration.)
• "I was not sure how specific to be." (They need the framework reinforced.)
• "I tried something it could not do." (Perfect -- this teaches boundaries.)
• "It was faster than I expected." (They are starting to see the value.)

Celebrate specific wins. If someone says "I used it to draft a client proposal and it saved me an hour," make a big deal of that. Real results motivate continued use.

Power Move 1: Custom Instructions and Personas (15 minutes)

Facilitator: Show this on screen. Create a custom instruction live.

Instead of giving context every single time, you can set up persistent instructions that shape every conversation. Think of it as training the AI to understand your world before you even ask a question.

Live Demo:

Show how to set up custom instructions:

1. "I work in [industry]. My clients are [description]. My communication style is [style]."
2. "When I ask for written content, default to [tone] and keep responses under [length] unless I say otherwise."
3. "Always consider [specific constraint or priority relevant to their work]."

Now show how the same prompt produces different results with and without these instructions. The difference is usually dramatic.

Power Move 2: Working with Data and Spreadsheets (15 minutes)

Facilitator: Prepare a sample spreadsheet beforehand. Nothing complex -- 20-30 rows of data with a few columns.

Most people do not realize they can upload a spreadsheet and have a conversation about the data. This is where ChatLLM becomes genuinely transformative for anyone who works with numbers.

Live Demo:

1. Upload a sample CSV or Excel file
2. Ask: "What does this data show? Summarize the key trends."
3. Ask: "Which rows are outliers and why?"
4. Ask: "Create a summary table grouped by [column name]."
5. Ask: "If I had to present this data to my boss in 3 bullet points, what would they be?"
6. Ask: "What questions should I be asking about this data that I have not asked yet?"

That last question is the sleeper hit. AI can identify angles you might not have considered. It is like having an analyst who asks "have you thought about this?"

Power Move 3: Chaining Tasks Into Workflows (15 minutes)

Facilitator: This is about thinking in sequences, not single prompts. Walk through a real multi-step workflow.

The biggest leap in AI productivity happens when you stop thinking about one task at a time and start thinking about sequences.

Example workflow for a weekly client report:

1. "Here is the raw data from this week. Identify the top 3 metrics that changed significantly."
2. "For each metric, explain what happened and why it might matter."
3. "Write a 200-word executive summary suitable for a non-technical client."
4. "Create 3 recommended action items based on the data."
5. "Draft an email to the client with the summary and recommendations attached. Tone: professional but approachable."

Five prompts, one continuous conversation. What used to take 2-3 hours now takes 20 minutes. And the output is often more thorough because the AI catches things you might miss when you are rushing.

Power Move 4: Using AI to Challenge Your Own Thinking (10 minutes)

This is an underused technique that separates good AI users from great ones.

The Devil's Advocate Prompt:

"I am planning to [describe your plan or decision]. Act as a skeptical advisor. What are the top 5 risks or weaknesses in this plan? Be honest and specific -- do not just validate my idea."

The Pre-Mortem Prompt:

"Imagine this project failed completely 6 months from now. What are the most likely reasons it failed? Work backwards from the failure to identify what I should be worried about now."

The Blind Spot Prompt:

"Based on everything I have told you about my business/project/plan, what am I probably not thinking about? What assumptions am I making that could be wrong?"

These prompts turn AI from a yes-man into a genuine thinking partner.

Hands-On Practice (15 minutes)

Facilitator: Let them choose what to work on. Offer guidance but let them drive.

Choose one:

1. Data Workout: Upload a real spreadsheet and have a full analytical conversation with ChatLLM. Ask at least 5 questions about the data.
2. Workflow Build: Take a multi-step task from your work and chain it into a 4-5 prompt conversation that produces a complete deliverable.
3. Strategic Challenge: Use the devil's advocate, pre-mortem, or blind spot prompt on a real decision you are facing.

Session 3 Takeaways

• Custom instructions eliminate repetitive context-setting. Set them up once, benefit every time.
• Spreadsheets and data analysis in ChatLLM are a massive time-saver. Upload and ask questions in plain language.
• Think in workflows, not single prompts. Chain tasks together for complete deliverables.
• Use AI to challenge your thinking, not just confirm it. The devil's advocate prompts are some of the most valuable tools you have.
• The more you use these tools, the more you discover what they can do. Keep experimenting.

Before Next Session

Next session, we enter the workshop. We are going to use AI Agent to build something real. To prepare:

1. Think of something you wish existed -- a simple tool, a website, a calculator, a form, a dashboard. Nothing complex. Something that would make your life easier.
2. Write down what it would do in 2-3 sentences. Do not worry about how to build it. Just describe what it would do.
3. Bring that description to Session 4. We are going to build it.