The AI Readiness App course · Intermediate · about 90 minutes · Free

AI for software and coding work

Use AI coding assistants to read, write, test and review code faster, while keeping correctness, security and ownership with you.

For: Software developers, engineers, data engineers, testers and anyone who writes scripts or code.

1. Lesson 1

   Understand code before changing it

   Get oriented in unfamiliar code quickly.

   1. Ask the assistant to explain a file or function: purpose, inputs, outputs, side effects.
   2. Ask where it is called from and what could break if it changes.
   3. Confirm the explanation by reading the code and running it.

   Try this at work: Pick an unfamiliar module in your codebase. Ask an AI coding assistant (for example GitHub Copilot, Claude Code or Cursor) to explain it and list its callers, then verify two claims by reading the code.

   How to check it: Both claims you checked are correct.

   Watch out: Assistants can describe code that is not there. Trust the code, not the summary.

2. Lesson 2

   Write code with tests first

   Get correct code, not just plausible code.

   1. Describe the behaviour and edge cases, and ask for tests first.
   2. Review the tests: do they check the right thing?
   3. Then ask for the code and run the tests.
   4. Ask for a second pass on error handling and edge cases.

   Try this at work: For your next small feature or fix, ask the assistant to write failing tests first, review them, then generate the code and make the tests pass.

   How to check it: Tests fail before the change and pass after, and cover the edge cases you named.

   Watch out: Generated tests can test the wrong thing and pass anyway. Read every assertion.

3. Lesson 3

   Review AI code for security

   Stop AI from adding vulnerabilities.

   1. Check input validation, authentication, authorisation and secrets handling.
   2. Check every new dependency exists and is the package you meant.
   3. Run your linters, type checks and security scanners.

   Try this at work: Review one AI-generated change against a checklist: input validation, access checks, secrets, new dependencies, error messages.

   How to check it: No secrets in code, every input validated, every dependency verified.

   Watch out: AI can suggest packages with names that look right but do not exist or are malicious. Verify the exact package name.

4. Lesson 4

   Automate the boring parts

   Use AI for scripts, migrations and documentation.

   1. Ask for scripts for repetitive tasks, with a dry-run mode.
   2. Ask for documentation drafts from the code and tests.
   3. Keep humans reviewing every merge.

   Try this at work: Ask the assistant to write a small script for a repetitive task you do, with a dry-run option, and test it on a copy of the data.

   How to check it: The dry run shows exactly what will change, and the real run matches it.

   Watch out: Never run AI-written scripts against production data before testing on a copy.

All courses