HOW AI TESTS SOFTWARETHE COURSE / JASON ARBON
The full course. at launch.

Planned regular price: $94. Save 50%.
Want personal guidance? Choose email Q&A or one 20-minute 1-on-1 Zoom.

Compare the three options ↓

Parts One–Three recorded · Free previews

How AI
Tests Software.

Turn AI into your testing partner.
Find the problems. Verify the evidence.
Make the next release better.

Explore the course

With Jason ArbonFor testers, developers, and QA leaders.

LEARN BY DOING. WITH JASON. An instructor helps a class turn a question into a test and evidence A QUESTION → EVIDENCE ObserveTestVerify
01 Observe02 Verify03 Explain

Short lessons. Concrete examples. Practice with your own AI agent.

56recorded lessons
3h 22mParts One–Three video
6free previews
Your AI agentportable methods, not one vendor
PART ONE / AI CHECKS

Find more.
Assume less.

Go beyond asking AI to “find bugs.” Learn to give it the right evidence, ask sharper questions, and challenge what comes back.

  1. 01

    Inspect more than the screen

    Use screenshots, console logs, network traces, code, and data to investigate a problem.

  2. 02

    Ask the question that matters

    Bring product context and focused checks to security, accessibility, performance, and more.

  3. 03

    Turn findings into useful work

    Challenge assumptions, merge duplicates, and write evidence-backed bugs that help someone fix them.

TAKE A LOOK INSIDE

A few lessons. On us.

Real excerpts from the recorded course.
No sign-in. No payment.

The introduction

Who the course is for, what you will learn, and where the six-part journey is headed.

The first AI-checks lesson

Inspect screenshots, logs, and source artifacts. Learn what a screenshot can and cannot prove.

A lesson from the middle

Turn known failure modes into reusable checking questions instead of one-off hard-coded cases.

PART 2 · FREE PREVIEW

See an AI coding agent interact with a browser, then verify the drink, quantity, and price in TestBucks.

PART 3 · FREE PREVIEW

Give your testing agent source code and business documents so it can reason about boundaries, dependencies, and product risks.

PART 3 · FREE PREVIEW

Explore running models locally with Ollama, including offline use and the tradeoffs of local hardware.

THE RECORDED CURRICULUM

Parts One–Three

56 lessons · 3 hours 22 minutes
AI checks, interactive tests, and contextual testing.

56 lessons

01Start here

  1. 01Welcome to How AI Tests Software3:41Free preview

    Who the course is for, what you will learn, and where the six-part journey is headed.

  2. 02Set up your AI coding agent4:25Recorded

    Get an agent reading local files, set safe boundaries, and choose a suitable practice environment.

    Full lesson access is not open yet.
  3. 03Meet TestBucks. Find your first bugs.1:49Recorded

    Use a fictional coffee shop to make your first evidence-backed observations.

    Full lesson access is not open yet.
  4. 04Checks, tests, harnesses, and confidence5:35Recorded

    Separate artifact inspection from interactive testing and longer-running quality workflows.

    Full lesson access is not open yet.

02Evidence and findings

  1. 05AI checks: start with the evidence4:54Free preview

    Inspect screenshots, logs, and source artifacts. Learn what a screenshot can and cannot prove.

  2. 06Write a bug report that helps fix the bug4:56Recorded

    Use OpenTest.ai-style fields for reasoning, counterarguments, routing, and a fix prompt.

    Full lesson access is not open yet.
  3. 07Find issues in logs and recorded evidence6:06Recorded

    Inspect network traces, console logs, and database artifacts, not just the visible page.

    Full lesson access is not open yet.
  4. 08Connect findings to your existing tools2:20Recorded

    Connect issue and test repositories while keeping a human review step before filing.

    Full lesson access is not open yet.

03Better questions, better checks

  1. 09Give your agent useful context5:15Recorded

    Combine product rules, code, logs, and team knowledge to investigate more effectively.

    Full lesson access is not open yet.
  2. 10Many pages. Fewer duplicate reports.6:47Recorded

    Find inconsistencies across pages and consolidate repeated findings into actionable issues.

    Full lesson access is not open yet.
  3. 11Ask focused questions2:36Free preview

    Turn known failure modes into reusable checking questions instead of one-off hard-coded cases.

  4. 12When the assistant refuses5:03Recorded

    Understand refusals, authorized testing context, and the product behavior that follows a refusal.

    Full lesson access is not open yet.

04Specialist checks

  1. 13Security and privacy checks1:58Recorded

    Look for cross-customer data, unintended disclosures, and inconsistencies in captured evidence.

    Full lesson access is not open yet.
  2. 14Usability and accessibility4:14Recorded

    Examine confusing choices, meaningful alternative text, zoom behavior, and visual accessibility.

    Full lesson access is not open yet.
  3. 15Performance, visual quality, and experience3:35Recorded

    Investigate heavy assets, awkward layouts, and experiences that undermine customer trust.

    Full lesson access is not open yet.
  4. 16Functionality, reliability, and compatibility5:42Recorded

    Check calculations, hidden retries, network problems, and responsive layouts.

    Full lesson access is not open yet.
  5. 17Content, localization, data, and APIs4:46Recorded

    Compare claims, locales, stored data, and API behavior for contradictions.

    Full lesson access is not open yet.
  6. 18Review AI-generated code2:53Recorded

    Ask for focused code review and challenge what the generating agent overlooked.

    Full lesson access is not open yet.
  7. 19Standards and compliance checks4:36Recorded

    Use AI to flag potential issues for expert review, not to certify legal compliance.

    Full lesson access is not open yet.

05Make the findings useful

  1. 20Reduce noise. Test state and relevance.3:22Recorded

    Bring known issues into the prompt, vary application state, and judge search relevance.

    Full lesson access is not open yet.
  2. 21Recovery, specifications, and parallel checks2:59Recorded

    Review recovery paths and specifications, then combine complementary artifact checks.

    Full lesson access is not open yet.
  3. 22Part One: the quality-checking loop2:30Recorded

    Connect evidence, focus the questions, challenge findings, and make bugs actionable.

    Full lesson access is not open yet.

02Part 2: Interactive AI Tests

  1. 2.01Interactive tests: act and verify3:11Recorded

    Interactive tests: act and verify. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  2. 2.02Let the agent drive the browser4:29Free preview

    See an AI coding agent interact with a browser, then verify the drink, quantity, and price in TestBucks.

  3. 2.03Connect your existing testing tools3:37Recorded

    Connect your existing testing tools. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  4. 2.04Use frameworks and save results to your tools2:58Recorded

    Use frameworks and save results to your tools. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  5. 2.05Adaptive tests and AI-first bug tracking3:40Recorded

    Adaptive tests and AI-first bug tracking. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  6. 2.06Why generated test code still breaks3:41Recorded

    Why generated test code still breaks. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  7. 2.07Save the intent, not the brittle path2:43Recorded

    Save the intent, not the brittle path. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  8. 2.08Executable intents and the OpenTest.ai schema4:38Recorded

    Executable intents and the OpenTest.ai schema. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  9. 2.09Preserve evidence and explain the result3:23Recorded

    Preserve evidence and explain the result. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  10. 2.10Turn evidence into a quality decision2:55Recorded

    Turn evidence into a quality decision. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  11. 2.11Part Two recap: interactive AI testing0:38Recorded

    Part Two recap: interactive AI testing. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.

03Part 3: Contextual Testing

  1. 3.01Give your agent source and business context5:07Free preview

    Give your testing agent source code and business documents so it can reason about boundaries, dependencies, and product risks.

  2. 3.02Challenge the implementation and its boundaries3:41Recorded

    Challenge the implementation and its boundaries. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  3. 3.03Prioritize risks with source-aware analysis3:49Recorded

    Prioritize risks with source-aware analysis. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  4. 3.04Formal verification and Lean: where AI helps3:25Recorded

    Formal verification and Lean: where AI helps. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  5. 3.05Check a focused property and its counterexamples4:07Recorded

    Check a focused property and its counterexamples. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  6. 3.06The changing role of the tester3:03Recorded

    The changing role of the tester. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  7. 3.07People and AI: review complementary findings4:18Recorded

    People and AI: review complementary findings. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  8. 3.08Generate test data and combinations at scale3:41Recorded

    Generate test data and combinations at scale. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  9. 3.09Manage review overload with sampling4:45Recorded

    Manage review overload with sampling. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  10. 3.10Balance model cost, capability, and speed6:00Recorded

    Balance model cost, capability, and speed. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  11. 3.11Run local models with Ollama1:18Free preview

    Explore running models locally with Ollama, including offline use and the tradeoffs of local hardware.

  12. 3.12Local models still have biases and risks2:15Recorded

    Local models still have biases and risks. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  13. 3.13Fine-tune a model on your triage examples3:14Recorded

    Fine-tune a model on your triage examples. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  14. 3.14Compare cost, parallelism, and availability2:52Recorded

    Compare cost, parallelism, and availability. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  15. 3.15Personas and skeptical review3:35Recorded

    Personas and skeptical review. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  16. 3.16Challenge your assumptions about AI testing3:06Recorded

    Challenge your assumptions about AI testing. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  17. 3.17Different paths, stable outcomes3:38Recorded

    Different paths, stable outcomes. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  18. 3.18Check meaning across varying responses2:15Recorded

    Check meaning across varying responses. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  19. 3.19Temperature, top-k, and top-p3:05Recorded

    Temperature, top-k, and top-p. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  20. 3.20Judge meaning with evidence2:40Recorded

    Judge meaning with evidence. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  21. 3.21Compare and calibrate independent judges1:25Recorded

    Compare and calibrate independent judges. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  22. 3.22Judge against evidence, not authority3:12Recorded

    Judge against evidence, not authority. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.
  23. 3.23Part Three recap: contextual testing1:24Recorded

    Part Three recap: contextual testing. A recorded TestBucks-based lesson with Jason Arbon.

    Full lesson access is not open yet.

Practice time is additional. The full lessons are recorded; enrollment is not open yet. Only the six marked previews are publicly playable.

THE REST OF THE JOURNEY

What comes next.

Planned topics for Parts Four through Six.
These videos have not been recorded yet.

04

Tests Become Intent

Not yet recorded
  • Keep the promise. Adapt the path.
  • Reuse intent across surfaces
  • Record a journey. Replay its intent.
  • Turn legacy tests into reusable intent
  • Make outcomes observable
  • Plan a bounded load test
  • Inject a fault. Verify recovery.
05

The AI Test Harness

Not yet recorded
  • Coordinate the testing workflow
  • Give specialists context and direction
  • Save evidence. Compare runs.
  • Build a release recommendation
  • Run isolated workers in parallel
  • Choose and steer the next test
06

The Future of Testing

Not yet recorded
  • Test changing experiences
  • Prove a focused property
  • Software that tests and repairs itself
  • Keep autonomous testing accountable
THREE OPTIONS · TOTAL PRICES IN USD

Choose how you want to learn.

Every option includes the full course. Pick email guidance or a single conversation if you want more.

OPTION 1

Full course

$47total at launch

Planned regular price: $94 Save 50%

  • All six course parts, as released
  • Slides and hands-on exercises

Learn at your own pace.

No enrollment cap

Choose this option
OPTION 2

Course + email Q&A

$97total at launch

Planned regular price: $144 Save $47

  • Everything in the full course
  • Three personal email responses from Jason

Ask about your questions and context. No Zoom session.

50 packages at launch

Choose this option
OPTION 3

Course + 1-on-1 Zoom

$247total at launch

Planned regular price: $297 Save $50

  • Everything in the full course
  • One 20-minute 1-on-1 Zoom with Jason

Discuss the contents of the book. Email Q&A is not included.

20 packages at launch

Choose this option

Free signup, no payment today. Select your preferred package in the form. This records your interest; it does not reserve a limited place or enroll you. Limited places are secured at paid enrollment. Prices above are complete package totals, not add-on charges.

Parts One–Three are recorded. Parts Four–Six will become available as they are released; they are not yet recorded. The launch date will be announced later.

Personal-support details

Email Q&A: Three responses in total, including follow-ups—not three ongoing threads. Use within 90 days of enrollment. Replies within five business days. One focused topic per email with brief context. No codebase audits, extensive attachment review, or ongoing troubleshooting.

1-on-1 Zoom: One 20-minute session to discuss the contents of How AI Tests Software. Book within 90 days of enrollment, subject to calendar availability. No email Q&A, advance code review, or written consulting report.

The 50 email packages and 20 Zoom packages are launch allocations, not live remaining-place counts.

Buy How AI Tests Software by Jason Arbon on Amazon
OPTIONAL COMPANION READING

The course comes first.
The book goes deeper.

Follow the lessons and practice the techniques here. Keep the book nearby when you want a longer explanation or a reference to return to.

Buy the book on Amazon →
Jason Arbon, author and course instructorYOUR INSTRUCTOR
TESTING EXPERIENCE, SHARED

Jason Arbon

“My personal mission is to improve the quality of the world's apps.”

Author of How AI Tests Software. In this course, Jason walks through the methods he uses to investigate software with AI, using the fictional TestBucks coffee shop as a shared example.

Watch the reasoning, try the checks, and question the results. The goal is useful testing, not simply more AI output.

Explore the companion book
TOOLS & SERVICES

Put the learning to work.

Codex

BY OPENAI

Your AI coding partner.

Bring the course into your own project. Use Codex to explore code, write tests, investigate bugs, and work through fixes.

Explore Codex

Iceberg QA

Find what your AI missed.

Add instant-on AI QA bandwidth and coverage, with human testing expertise. Start getting value in days, and get help learning to run it yourself.

Talk to Iceberg QA
START WITH ONE GOOD QUESTION

What did your testing miss?

See how a focused AI check can help you find out.

Register to receive details about your selected course launch offer from Jason. Name and email are required; workplace, job title, and learning goals are optional. Other AI updates require a separate opt-in. Responses are stored privately in Google Sheets. Please do not include confidential information. Withdraw by emailing jason@testers.ai. The launch date will be announced later. Prices are in USD. Visit the book homepage →

FREE COURSE PREVIEW

Read the transcript
Part One / AI Checks

Get your launch offer

No payment today. Choose your package and I’ll email enrollment details.