AI-assisted work with engineering judgment
Opening notes
Welcome the viewer. Explain that this is one lesson in the eight-part Job-Ready Software Engineering series. Set up the outcome before beginning.
Speaker view / 07
AI-assisted work with engineering judgment
Faster is not the same as better
Use the tool. Own the result.
Speaker notes
AI can reduce the cost of a first draft. It does not remove the responsibility to understand, test, secure, and explain the result.
Good uses for AI
- Explain unfamiliar code
- Propose test cases
- Draft documentation
- Explore alternatives
- Generate a bounded first draft
Speaker notes
Choose tasks where a human can verify the output. AI is most useful for bounded tasks with clear inputs and observable results.
A useful task prompt
- Context
- Task
- Constraints
- Output format
Speaker notes
Use a Launchpad example: propose API tests for creating an application, do not change code, include negative cases, and return a table with risks and expected responses.
The verification loop
- Ask
- Inspect
- Run
- Compare
- Correct
- Document
Speaker notes
Never treat a plausible answer as a verified answer. Run tests, inspect assumptions, compare with documentation, and ask what could be wrong.
What must not be shared
- Passwords and tokens
- Private customer data
- Employer code without permission
- Sensitive logs
- Confidential plans
Speaker notes
Use a safe sample project for demonstrations. Follow employer policy when an AI tool is approved.
Demo: review an AI-generated patch
- Correctness
- Edge cases
- Security
- Test coverage
- Maintainability
- Scope
Speaker notes
Ask an AI tool for a small Launchpad change. Run tests, find a missed case, correct the patch, and record what changed.
Pause: create an evidence log
- Prompt
- Output
- Verification steps
- Mistakes found
- Final correction
Speaker notes
Pause the video. The evidence log is more valuable than claiming that AI built something. It shows judgment.
Your artifact
I used AI to accelerate __, verified it by __, found __, and changed __ before shipping.
Speaker notes
That sentence becomes an interview answer. Employers need to hear how viewers supervise tools.
Make the artifact.
Publish your work, then continue to the next lesson.
Closing notes
Repeat the artifact viewers should create. Invite them to subscribe and join the future live cohort waitlist.