You’re not slow. You start from zero every time.

When a role wants an agentic prototype by afternoon, it is not testing how fast you type. It is testing whether you still rebuild the same scaffolding every time: project setup, an auth stub, a streaming backend, a trace view, a review gate. Spending those hours again is not a skill gap. It is a tooling gap. Tooling is something you can learn to build once and reuse. In this course you will build that kit, then practice under a fixed clock so you can prove the claim, not just assert it.

Start lesson one: Where the hours actually go →

Lessons

This course is being written as you work through it. The first module is ready. It decides what the kit is allowed to contain, which you must settle before there is anything worth building. Ask your teaching agent for the next module once your first timed run is in the log.

How to actually get value from this

• Time the first run before you improve anything. The course opens with a stopwatch, not a template. Later cuts only make sense if they come from that measurement. Skip the first timing and you will build a kit around a guess.
• Judge the kit by what it leaves out. If an entry encodes a decision the next prototype should still get to make, it is a liability dressed up as a head start.
• Assume the kit is decaying while you read this. The usual failure mode is not a bad first design. It is six months of drift, then an hour spent discovering what broke before you can start the real work.