Skip to content

All courses

Course 10 of The AI Decision Course

AI for software and technical teams

For leaders whose software teams are picking up AI coding tools: what to measure before and after, who owns which work, review and gates at agent speed, and what the tools may see.

11 short lessons. Each one is a short read with a card your team fills in and three AI prompts to test it. Part of The AI Decision Course: membership is $299 a year and includes all 12 courses, with a full refund within 14 days.

Join for $299 a yearRead the free lesson

About this course

This course covers bringing AI coding tools into a software organization: the delivery numbers to watch before the tools land, which coding work goes to AI and who owns the rule, and funding what the code reads before it writes. It also covers review and release gates that hold at agent speed, what a coding agent may do alone in your repo, and classifying repositories before the tools arrive.

You leave able to set a gate every AI-written change must pass and to size the review line before generated code floods it. You can also classify repositories by what they hold, keep a person on merge and deploy, and set a delivery rhythm that survives model churn.

Lessons in this course

  1. Pick the delivery numbers to watch before coding tools land3-minute read, Free to read
  2. Set which coding work goes to AI, and who owns the rule2-minute read
  3. Before AI writes code, fund what it reads3-minute read
  4. When code is cheap, the spec becomes the bottleneck3-minute read
  5. When AI writes ten times the code, size the review line2-minute read
  6. Set the gate every AI-written change must pass3-minute read
  7. What a coding agent may do alone in your repo3-minute read
  8. Set a delivery rhythm that survives the churn2-minute read
  9. One paved road for AI coding tools, or each team its own3-minute read
  10. Before the coding tools arrive, classify what they may see3-minute read
  11. Own the long tail of internal tools your engineers build3-minute read

Back to the course list

Questions

How do we measure the impact of AI coding tools?

Pull throughput, stability and review load for your recent quarters while the before picture still exists, so the net effect shows in your own numbers. Watch the whole delivery cycle rather than coding time alone, and treat review as the next bottleneck. The lesson 'Pick the delivery numbers to watch before coding tools land' covers the measurement.

How should we review AI-generated code?

Plan review capacity before the code volume arrives, and count errors shipped rather than error rates. Reviewers look for human-style mistakes and miss the unfamiliar ones AI makes, so keep the corner cases and the spec with a human, and have developers modify critical generated code until the team stands behind it. The lesson 'When AI writes ten times the code, size the review line' covers the sizing.

What should AI coding tools be allowed to see?

Classify your repositories before the tools arrive, from public to regulated, and let the class set the tool tier and the contract terms. The business names what is proprietary, regulated data stays inside your own boundary, and a person reviews changes to the core repository. The lesson 'Before the coding tools arrive, classify what they may see' covers the classes.

The other courses in the membership