
Course Content
Facilitator Guide: Session 4
Duration: 90 minutes. This is the "wow" session. When people see AI Agent build a working website from a conversation, it fundamentally shifts their understanding of what is possible. Take your time. Let the magic land.
Opening: The Mindset Shift (10 minutes)
Ask: "What did you come up with? What is the thing you wish existed?"
Listen to their idea. Then say this: "By the end of this session, that thing is going to exist. Not a sketch. Not a plan. A real, working thing you can use and share."
Let that sink in. This is the moment that changes everything for most people.
ChatLLM vs. AI Agent: The Key Difference (10 minutes)
Facilitator: Use an analogy they can relate to.
Here is the simplest way to understand the difference:
ChatLLM is like talking to a consultant. You describe your problem, they analyze it, they give you advice, they write documents, they help you think. But at the end of the conversation, you still have to go execute.
AI Agent is like hiring a contractor. You describe what you want built, they ask clarifying questions, and then they build it. They write the code, create the files, set up the database, deploy the website. You describe, they deliver.
Both are valuable. A consultant helps you think. A contractor helps you build. AI Agent is the contractor.
Navigating AI Agent: The Interface (10 minutes)
Facilitator: Share screen. Open AI Agent in Abacus AI. Walk through the interface slowly.
Key areas to point out:
- The conversation panel. This is where you talk to AI Agent. It looks similar to ChatLLM, but the AI here is designed to take action, not just respond.
- The preview panel. As AI Agent builds, you see the result in real time. This is the screen where your website or app appears.
- The file explorer. AI Agent creates real files -- code, images, configurations. You can see them here. You do not need to understand them, but it is good to know they exist.
- Deploy button. When you are happy with what is built, one click makes it live on the internet with a real URL anyone can visit.
Important to emphasize: You are not learning to code. You are learning to communicate clearly about what you want. AI Agent handles the technical execution. Your job is to be a great communicator and a quality checker.
Live Demo: Building Something from Scratch (20 minutes)
Facilitator: Build something live. Choose something relevant to the participant's world. Type slowly enough that they can follow. Narrate your thinking.
Suggested demo project -- a Client Intake Form:
Type into AI Agent (show the actual prompt):
"Build me a clean, professional client intake form for a consulting business. It should collect: full name, email address, phone number, company name, company size (dropdown: 1-10, 11-50, 51-200, 200+), a brief description of what they need help with (text area), and their preferred contact method (radio buttons: email, phone, video call). The design should be modern, mobile-friendly, and use a navy and gold color scheme. When someone submits the form, show them a confirmation message that says their information has been received and someone will be in touch within 24 hours."
As it builds, narrate:
- "Notice how it is creating the form fields I described."
- "See the color scheme matching what I asked for."
- "It is making it mobile-friendly automatically because I asked."
- "Now watch -- it is adding the confirmation message."
Then refine:
- "Make the submit button larger and add the text 'Get Started' instead of 'Submit.'"
- "Add a brief paragraph above the form explaining why they should fill it out."
- "The phone number field should auto-format as they type."
Show that iteration works the same way here as in ChatLLM. You describe what you want changed, and it changes.
The Art of Clear Instructions for AI Agent (10 minutes)
AI Agent responds to the same principles as ChatLLM, but with some additions:
- Describe the end result, not the process. Say "I want a page that shows a list of team members with their photos and titles" not "write HTML code for a grid layout with image tags." You describe the what, AI Agent figures out the how.
- Be specific about design. Colors, layout, fonts, mood. "Modern and clean" is better than nothing, but "Navy background with white text, rounded corners on cards, generous spacing between sections" is better.
- Describe user interaction. What should happen when someone clicks a button? What should the form do when submitted? What should the page look like on a phone vs. a computer?
- Iterate in small steps. Do not try to describe everything in one giant message. Build the foundation first, then add features one at a time. This gives you control and makes it easier to catch problems early.
- Check as you go. Look at the preview after each change. If something is not right, say so immediately. Do not wait until 10 changes later to mention the first one was wrong.
Your Turn: Build Your Idea (25 minutes)
Facilitator: This is the main event. Guide them through building the thing they described in their homework. Help them write their first prompt. Coach them through iteration.
Step 1: Write your opening prompt (5 minutes)
Using the description you brought, write a clear, detailed prompt for AI Agent. Include:
- What it is (a form, a website, a dashboard, a calculator)
- Who it is for (you, your clients, your team)
- What it should look like (colors, style, mood)
- What users should be able to do
Step 2: Watch it build (5 minutes)
Submit your prompt and watch AI Agent work. Do not interrupt -- let it finish the first version.
Step 3: Review and refine (10 minutes)
Look at what it built. What is right? What needs adjustment? Send feedback as clear, specific instructions.
Step 4: Polish (5 minutes)
Make 2-3 more refinements until it feels right.
Deploying: Making It Real (5 minutes)
Facilitator: Show the deploy process. This is the moment it stops being a demo and starts being real.
When you are happy with what you built:
- Click the Deploy button
- Your project gets a real URL -- a live website anyone in the world can visit
- Share that URL with a colleague, a friend, or yourself on your phone
- Take a moment to appreciate what just happened: you described something, and a working tool now exists on the internet
Session 4 Takeaways
- AI Agent builds things. Real, working websites, forms, tools, and applications.
- Your job is to communicate clearly and check quality. AI Agent handles the technical work.
- Describe the end result, not the process. Focus on what it should do, not how to code it.
- Iterate in small steps. Build, review, refine, repeat.
- Deploy makes it real. One click and your creation is live on the internet.
Before Next Session
- Share what you built with at least one person. Get their reaction.
- Think about what you would add or change. Write down 3-5 improvements.
- Think bigger: if AI Agent can build that in 30 minutes, what else could it build for your business?
