A series by Richard Bumann
Let your inner child drive.
One app a week, built with AI. And the funniest, strangest, most useful conversations along the way.
The tower grows by one block every week. Drag to turn it, click a block or a speech bubble to open it.
Strand 1
One App a Week
Each week, one app I built for learning or work, with pro-code and low-code tools, told as a short interactive story (about 3 minutes, narrated): the trigger, the idea, the architecture, the learnings and the products behind it.
Strand 2
Conversations with AI
Under a minute each. Funny, bizarre, but useful moments from my chats with Copilot, Cowork, desktop agents and GitHub Copilot, re-enacted with synthetic voices, with the punchline in print.
Why this series
Play. Ask why. Try again.
Toddlers don't read the manual: they try, fail, laugh and try again. That is the mindset that makes AI work. This series shares what I build and what I learn, in the open, one small step at a time.
Every episode passes a strict confidentiality check (content and code) before it is published. Data on screen is fictional or public. Voices are synthetic.
- Spec, design, buildSpec first, then design, then iterate, with AI at every step.
- Precise vs. openPin down what you know. Invite creativity where you want new options.
- Experiment smallLet AI test ideas on small samples before you scale.
- Engineering rigorVersions, reviews and tests for AI workflows too, not just code.
- Security built inSecurity and confidentiality checks in every loop, not at the end.
- Speed commoditizesToday's edge is tomorrow's feature. Invest in the problem and the know-how.