FREE · ENGLISH · NO CODING REQUIRED
Build your first AI + QA test pack
Five written exercises you can complete now, at your own pace. Allow 15–25 minutes per exercise. An AI chat tool is optional; you can do every exercise manually. Video lessons are planned, not yet available.
Save your work in a document or repository. These public exercises do not track completion or issue a certificate. Your deliverable is a reviewed test pack you can explain.
Shared practice brief: the demo checkout
This is a fictional exercise, not a customer incident. Use the same brief for all five lessons.
- One product costs $20 per unit. Quantity must be an integer from 1 to 10 inclusive.
- Empty coupon means no discount. SAVE10 applies a 10% discount to the subtotal.
- Coupon matching is case-sensitive. Reject any other non-empty coupon.
- Shipping, tax, stock, and payment rules are outside this brief; ask for clarification.
- Fictional defect: quantity 1 with SAVE10 displays $20 rather than $18.
- Fictional run results: quantity boundary checks pass; the SAVE10 check fails; another supported browser has not been tested. There is no deployment evidence.
EXERCISE 1 OF 5
Review requirements before generating tests
Separate confirmed rules from assumptions.
Try this prompt with the shared brief
Act as a QA reviewer. Using only the checkout rules below, list confirmed acceptance criteria and unanswered questions separately. Cite the rule behind every criterion. Do not invent shipping, tax, or payment behavior.
Your task: Write three clarification questions and five acceptance criteria.
Complete when: Every criterion traces to a stated rule; unknowns are labeled.
Compare with the worked example after your attempt
Confirmed: quantity must be an integer from 1 through 10; subtotal is quantity × $20; SAVE10 discounts subtotal by 10%; other non-empty coupon codes are rejected; coupon matching is case-sensitive. Unknown: shipping, tax, stock, and payment rules. Ask about them without adding them to scope.
EXERCISE 2 OF 5
Generate and review useful test cases
Cover boundaries and negative scenarios using AI as a draft assistant.
Try this prompt with the shared brief
Draft eight test cases from the supplied checkout rules. Include quantity boundaries, valid and invalid coupons, and expected totals. Return ID, rule, input, steps, expected result. Label anything that needs clarification. I will review every result.
Your task: Produce eight cases with input, steps, and expected result.
Complete when: Include valid, invalid, and boundary cases without unsupported rules.
Compare with the worked example after your attempt
Useful cases: qty 0 rejected; qty 1/no coupon = $20; qty 10/no coupon = $200; qty 11 rejected; qty 1/SAVE10 = $18; qty 10/SAVE10 = $180; qty 1/WRONG rejected; qty 1/save10 rejected. These cover a subset of the risks, not all possible checkout behavior.
EXERCISE 3 OF 5
Write a reproducible bug report
Turn an observation into evidence a teammate can reproduce.
Try this prompt with the shared brief
Turn the supplied fictional defect observation into a reproducible bug report. Include environment, exact steps, expected and actual result, impact, suggested severity with a reason, and evidence needed. Do not invent a screenshot or a reproduction attempt.
Your task: Write a report with environment, steps, expected/actual behavior, and impact.
Complete when: Someone else can reproduce it; severity has a reason.
Compare with the worked example after your attempt
Report: SAVE10 is ignored at quantity 1 in the fictional fixture. Environment: supplied exercise fixture (not a live production report). Steps: choose quantity 1, enter SAVE10, apply. Expected $18; observed $20. Impact: promised discount is not applied. Suggested priority: resolve before coupon launch; business owner confirms severity/priority. Evidence needed: actual screenshot and repeatable run when reproduced in a real test environment.
EXERCISE 4 OF 5
Build safe, purposeful test data
Design synthetic datasets with clear coverage.
Try this prompt with the shared brief
Generate ten synthetic checkout input rows using the supplied rules. Include valid, invalid, and boundary cases. Return quantity, coupon, expected validation or total, and the rule covered. Do not include personal data.
Your task: Create ten synthetic rows and classify each.
Complete when: No real customer data; each row has a rule and expected outcome.
Compare with the worked example after your attempt
Example extra rows: qty 1.5 should be rejected because quantity must be integer; qty -1 should be rejected; qty 2/SAVE10 = $36. For an omitted quantity, ask what default/error behavior is required rather than inventing it. Use no real customer records.
EXERCISE 5 OF 5
Make an evidence-based release recommendation
Distinguish a completed check from a release claim.
Try this prompt with the shared brief
Draft a release recommendation from only the supplied run results and known gaps. Separate verified evidence, remaining risk, exact decision needed, accountable role, and next action. Never treat an untested browser or a missing report as a pass.
Your task: Write a release recommendation with evidence, risk, owner, and next step.
Complete when: No invented pass results; missing coverage and decision owner are explicit.
Compare with the worked example after your attempt
Recommend holding the coupon feature release until the confirmed discount defect is fixed and retested. The browser coverage gap needs a risk decision. Request a named release decision-maker and a committed retest date; neither is supplied by the fixture. Do not invent them or announce deployment.
Review the complete pack
Check accuracy, traceability, reproducibility, and honest treatment of missing information. Revise any unsupported claim. Ask a peer to reproduce one case from your instructions.
