PUBLIC_KEY_PEM before distributing.
You've been using a standalone tool since Discover AI Module 3 — the Context Builder. Now you build one of your own. You describe a repetitive task, prompt an LLM to generate an HTML tool for it, and iterate until it's genuinely useful.
By the end of this module: you've identified a repetitive business task, generated a standalone HTML tool for it, iterated on that tool with a follow-up prompt, and reflected on what changed between versions.
The Context Builder you downloaded in Discover AI Module 3 is a standalone HTML file — a small, self-contained tool that runs entirely in your browser. No server. No login. No installation. Just a file you open and use.
That file was generated by an LLM. Not manually coded — described, prompted, and produced in a single conversation. That's what this module teaches: how to do that for yourself.
A single .html file
that contains everything — layout, logic, and any data — in one place.
Open it in any browser. It works offline. It saves or exports its own output.
No dependencies, no accounts, no network. This is what you ask the LLM to build.
The skill is in describing what you want clearly enough that the LLM can build it in one shot — or close to it. The same context-building skills you practiced in Discover AI Modules 1 and 2 apply here. Role, content, instructions. Except now the output is running code, not a written response.
Things you do repeatedly, with the same structure, but variable inputs. Formatting a weekly status update. Calculating project margins. Drafting a structured meeting agenda. Scoring a checklist. Anything where you currently use a blank document and fill in the same sections every time is a candidate.
Paste the broken code back into a new conversation and describe what isn't working. Or ask the LLM to review and fix it. The code is just text — the same iteration technique from Discover AI Module 4 applies. This is expected. First attempts rarely work perfectly. That's the point of Part 2.
Two parts. Part 1 is your first attempt — describe a task, write a prompt, generate the tool. Part 2 is iteration — write a follow-up prompt, generate an improved version, and note what changed.
Don't just ask "build me a tool." Use your Context Builder. Set a role (experienced web developer who builds clean, practical HTML tools for business users), describe what your tool needs to do in the content section, and write specific instructions about the interface, outputs, and constraints. The clearer your context, the closer to usable the first version will be.
Something you do regularly that has a consistent structure — same steps, different details each time. Write a description of it in the form below.
Use your Context Builder. Describe the tool's purpose, what inputs it takes, what it outputs or displays, and any constraints (standalone HTML, no internet, clean interface).
Save the code as a .html file and open it in your browser. See if it works as expected.
What didn't work? What could be better? Write a specific prompt addressing those issues and generate an improved version.
Paste the code from both versions and write brief reflections on each.
"Every Monday I write a weekly status update for my team. It always has the same four sections: what was completed last week, what's in progress, any blockers, and what's planned for next week. I want a tool where I can fill in each section, click a button, and get the whole thing formatted and ready to paste into an email or Slack message."
Two parts: your first attempt and your iterated version. All fields required.
Describe the repetitive task you identified. What is it, how often do you do it, and what structure does it follow?
Paste the full prompt (including role, context, and instructions if you used the Context Builder format).
Paste the complete HTML code the LLM produced. Include everything from <!DOCTYPE html> to </html>.
Did it work as expected? What was missing or off? What would you improve?
What did you ask the LLM to change or improve? Paste the prompt you used for the second version.
Paste the complete HTML code from the improved version.
Describe the specific improvements. Is this tool something you'd actually use?
Your work is saved locally. Download to keep a copy for offline reference. Your work has been saved. Keep this file for offline reference.