Session 2: ChatLLM -- Having Smart Conversations with AI

Facilitator Guide: Session 2

Duration: 75-90 minutes. This is the first real hands-on session. Expect questions, confusion, and breakthroughs. Go at their pace, not yours. The goal is confidence, not speed.

Opening: Check-In and Homework Review (10 minutes)

Ask: "Did you use AI at all since our last session? What did you try? What happened?"

Celebrate any attempt, even unsuccessful ones. If they did not try, no judgment -- ask what held them back. That information is valuable for how you pace the rest of the course.

What ChatLLM Actually Is (10 minutes)

Facilitator: Share screen showing ChatLLM interface.

ChatLLM is Abacus AI's conversation interface. Here is what makes it different from just going to chatgpt.com:

• Multiple AI models in one place. You are not locked into one AI. ChatLLM gives you access to different models -- think of them as different specialists. Some are better at creative writing, some at analysis, some at coding. You do not need to know which is which right now, but the option is there.
• Conversation memory. It remembers what you said earlier in the conversation. You can build on previous responses, refine them, go deeper. It is a real back-and-forth, not just question-and-answer.
• File handling. You can upload documents, spreadsheets, images, and PDFs. The AI can read them, analyze them, and work with the content. This is a game-changer for anyone who works with documents.
• Web search. ChatLLM can search the internet for current information. Regular AI models have a knowledge cutoff -- they do not know about things that happened after their training. ChatLLM with web search can find today's information.

The Anatomy of a Good Prompt (15 minutes)

Facilitator: This is the most important section. Use live examples. Type in real prompts and show the difference in results.

A prompt is just what you type into the AI. But not all prompts are created equal. Here is the difference:

Weak prompt: "Write me an email."

Result: A generic, useless email that could be about anything.

Strong prompt: "Write a professional email to a client named Sarah who missed our last meeting. The tone should be understanding, not passive-aggressive. Remind her about the meeting, suggest two new times next week (Tuesday at 2pm or Thursday at 10am), and mention that we have some exciting project updates to share. Keep it under 150 words."

Result: An email you could actually send.

The difference is specificity. Here is a framework you can use for any prompt:

1. Role: Who should the AI be? "Act as a business consultant" or "You are an experienced copywriter."
2. Task: What do you need done? "Write," "Analyze," "Summarize," "Compare," "Explain."
3. Context: What background information does the AI need? Industry, audience, purpose, constraints.
4. Format: How should the output look? Bullet points, paragraphs, table, specific word count.
5. Tone: How should it sound? Professional, casual, technical, friendly, urgent.

You do not need all five every time. But the more you include, the better your results will be.

Live Demo: Real Conversations (20 minutes)

Facilitator: Share screen. Do each of these live, typing in real time. Let the participant see your thought process.

Demo 1: The Back-and-Forth

Start with a broad request, then refine it through conversation:

1. "I need to create a presentation about our Q1 results for my team."
2. "Actually, make it focused on the three biggest wins and one area for improvement."
3. "The audience is non-technical. Remove any jargon."
4. "Add a suggested talking point for each slide."

Show how each follow-up makes the output better. This is how you work with AI -- iteratively, not in one shot.

Demo 2: Document Analysis

1. Upload a sample document (a report, article, or business plan)
2. Ask: "Summarize this document in 5 bullet points."
3. Follow up: "What are the three strongest arguments in this document?"
4. Follow up: "What is missing? What questions does this document not answer?"

This shows the participant that AI can be a thinking partner, not just a text generator.

Demo 3: Research with Web Search

1. Ask a question about a current topic in the participant's industry
2. Show how ChatLLM searches the web and synthesizes current information
3. Ask it to compare multiple sources or perspectives
4. Ask it to create an executive summary of its findings

Hands-On Practice (20 minutes)

Facilitator: Now it is their turn. Stay on the call. Watch their screen if possible. Coach in real time.

Exercise 1: The Email Rewrite

Take a real email you need to write (or recently wrote). Give ChatLLM the context and ask it to draft the email. Then:

• Ask it to make the tone more [formal/casual/direct/diplomatic]
• Ask it to shorten it by half
• Ask it to add a specific call to action

Exercise 2: The Document Deep Dive

Upload a document you are working with. Ask ChatLLM to:

• Summarize the key points
• Identify any action items
• Suggest improvements or flag potential issues
• Draft a response or follow-up communication based on the document

Exercise 3: The Research Sprint

Think of a question relevant to your work or industry. Ask ChatLLM to research it with web search enabled. Then ask follow-up questions to go deeper.

Common Mistakes and How to Avoid Them (10 minutes)

Facilitator: Share these from experience. Be honest about mistakes you have made too.

• Being too vague. "Help me with my business" gives you nothing useful. "Help me write a 30-day social media content calendar for a small accounting firm targeting small business owners" gives you gold.
• Accepting the first response. The first answer is a starting point, not the finish line. Always refine. Always ask for more.
• Not providing context. The AI does not know your industry, your audience, your goals, or your constraints unless you tell it. More context equals better output, every single time.
• Forgetting to verify. AI can be confidently wrong. If it gives you facts, statistics, or citations -- check them. Use AI for speed, use your brain for accuracy.
• Pasting confidential data without thinking. Before you paste anything into an AI tool, ask yourself: "Would I be comfortable if this information became public?" If the answer is no, redact sensitive details first.

Session 2 Takeaways

• ChatLLM is a conversation, not a search engine. The more you engage, the better the results.
• Good prompts have specificity: role, task, context, format, and tone.
• Always refine. Your first prompt is a starting point, not the final product.
• Upload documents to unlock AI as a thinking partner, not just a writer.
• Verify everything. Speed is the AI's job. Accuracy is yours.

Before Next Session

Use ChatLLM at least 3 times this week for real work tasks. Keep a brief note of:

1. What you asked
2. How useful the response was (1-10)
3. What you would do differently next time

Bring these notes to Session 3. We will use them to level up your skills.