The AI Edge · Watch · Read · MAKE
MAKE
Pick something you care about, or something you want to learn — then make it. Watch and read are inputs. MAKE is where the learning actually happens.
drC's approach
- Pick a medium — code, data, research, IoT, or media. Whatever you can ship.
- Use AI as a collaborator — Claude drafts, debugs, explains. You steer.
- Push it to GitHub — code, design files, writing, all of it, with a README and a screenshot.
- Loop when stuck — describe the problem to Claude and let it ask the clarifying questions.
Start with anything you can ship
No coding required. The point is to make something real with AI as your partner — and to learn by doing, not just reading.
Make something small
Pick a topic you care about and use AI to produce a finished thing — a short explainer, a planned trip, a research summary, a data chart, a piece of writing. Ship a v1, however rough.
Steer the collaboration
Treat Claude as a partner, not a vending machine. Ask it to draft, then push back, refine, and explain its choices. Notice how the result improves when you steer.
Loop when stuck
When you hit a wall, describe the problem plainly and let Claude ask you the clarifying questions. The back-and-forth is the skill — get comfortable with it.
MAKE for the Seven Steps
You've worked through the curriculum — now put each piece to use. Five hands-on exercises, one building on the last.
Step 1 · Command line
Use CLI tools to create a project directory and subdirectories. Use touch to create empty files, open nano to add some demo content, navigate back to the project root, and use grep to find that content across the files you made.
Steps 2–3 · Environments
Create two environments for the same project — one with venv + pip, another with uv. Compare the speed of creation and the ease of the commands. Feel the difference for yourself.
Step 6 · API keys & first LLM call
Get an API key from Anthropic and store it with dotenv. Then, inside an environment, write a Python program that uses the key to make a call to an Anthropic model and prints the response.
Step 5 · Ship it to GitHub
Create a GitHub account if you don't have one, install git on your machine, and commit your LLM-calling project to a public repo — with a README explaining what it does.
Step 7 · Local vs. frontier
Download a local model with Ollama and run the same prompt through it and through a frontier model. Compare the output quality and the speed. Note where each one wins.
Show drC what you made
Made something? Got stuck somewhere? Have an idea to make this site better? I want to hear about all of it — the wins and the walls.