Full 20-slide decks built for ~3 hours of lecture a week, then automatically trimmed to fit a quarter or an intensive summer. Click any slide to read it full-size.
The .pptx matches your selected term below, opens in PowerPoint and imports into Google Slides for editing.
Full 20-slide decks at ~3 hours of lecture a week, one chapter per week (a few weeks pair two), plus two optional-chapter weeks before final presentations.
~3 hrs/week · 14 weeksWeek 1
Week 2
Week 3
Week 4
Week 5
Week 6
Week 7
Week 8
Week 9
Week 10
Week 11
Week 12
Week 13
Week 14
Five multiple-choice questions drawn from the core chapters. Click an answer to check yourself, these are also included at the end of the slides PDF.
Q1. In the vibecoding loop (intent, generation, verification, refinement), which two stages belong to the human?
Q2. Why does a human's verification role become MORE important as AI tools gain agency, not less?
Q3. Which tool category lets the model read multiple files, propose multi-file diffs, AND execute commands inside your project?
Q4. A model confidently uses a function from a library version that doesn't exist. What's this an example of?
Q5. Why does the SAME underlying model sometimes feel much smarter in one tool than another?
A practical guide to making real software with AI.
Full decks for every chapter. Free for classroom and personal use under CC BY-NC 4.0.
Slide 1
Speaker note: Set expectations: this is a building course. Students ship things.
Slide 2
Slide 3
Speaker note: Draw this as a literal loop on the board. Every later chapter zooms into one arrow.
Slide 4
Slide 5
Icebreaker · 15 min · optionalSpeaker note: Facilitation: 5 min pair discussion, 5 min share-out, 5 min instructor synthesis tying answers to the loop on slide 3.
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Discussion · 15 min · optionalSpeaker note: Facilitation: small groups for 8 min, then 7 min whole-class share. Push students to name the actual trade-off (speed vs. control) instead of a tool name.
Slide 16
Slide 17
Slide 18
Slide 19
Slide 20
Slide 21
Slide 22
Slide 23
Slide 24
Slide 25
Quiz · 15 min · optionalSpeaker note: Use AlgoArena Classroom's live quiz mode. The close-split scenarios are usually where the real trade-off discussion lives, so spend debrief time there.
Slide 26
Slide 27
Slide 28
Slide 29
Speaker note: Definitions for the key terms in Chapter 1. Walk these before the recap; students can self-check against the chapter glossary.
Slide 30
Speaker note: Continued key-term definitions for Chapter 1. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
Demo · 15 min · optionalSpeaker note: Keep the prompt deliberately vague so the divergence is obvious. The punchline: repetition isn't a spec. Acceptance criteria are.
Slide 8
Icebreaker · 15 min · optionalSpeaker note: Keeps it light but makes the 'model fills gaps' lesson personal and memorable before the PRD content lands.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Opinion · 15 min · optionalSpeaker note: Use AlgoArena Classroom poll or a show-of-hands. Pick a PRD with at least one real, debatable gap (e.g. missing acceptance criteria) to get genuine disagreement.
Slide 14
Slide 15
Slide 16
Slide 17
Speaker note: This is the empirical backbone of the whole chapter: the PRD habit (2.2) and the reset habit (2.4) are the same finding seen from two ends. Source: Laban et al. 2025, in the chapter endnotes.
Slide 18
Slide 19
Slide 20
Design Challenge · 15 min · optionalSpeaker note: Bring a real example of the vague-vs-structured output difference if you have one from prep, to confirm the winning group's instinct.
Slide 21
Slide 22
Slide 23
Slide 24
Slide 25
Speaker note: Definitions for the key terms in Chapter 2. Walk these before the recap; students can self-check against the chapter glossary.
Slide 26
Speaker note: Continued key-term definitions for Chapter 2. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Icebreaker · 15 min · optionalSpeaker note: This works best after students have had at least one prior session with an AI IDE. Adjust if this is everyone's first exposure.
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Slide 11
Discussion · 15 min · optionalSpeaker note: If you have a real example from your own use of these tools, use it. Concrete beats hypothetical here.
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Quiz · 15 min · optionalSpeaker note: The disagreements usually reveal where students still conflate mechanical transformation with design judgment. That's the real teaching moment.
Slide 17
Slide 18
Slide 19
Speaker note: Tie this straight back to the steering slide. Students who accepted 'interrupt at step three' will feel the contradiction immediately, which is the point.
Slide 20
Slide 21
Speaker note: Anthropic's own multi-agent write-up says most coding tasks contain fewer genuinely parallel pieces than research does. Worth reading aloud.
Slide 22
Discussion · 10 min · optionalSpeaker note: Push students who say 'I'd review the diff anyway.' Ask how many lines they actually read on the last large diff they approved.
Slide 23
Slide 24
Slide 25
Slide 26
Speaker note: Definitions for the key terms in Chapter 3. Walk these before the recap; students can self-check against the chapter glossary.
Slide 27
Speaker note: Continued key-term definitions for Chapter 3. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Icebreaker · 15 min · optionalSpeaker note: Use any tokenizer playground to get the real count beforehand. The point is building intuition that a token is not a word.
Slide 5
Slide 6
Slide 7
Slide 8
Slide 9
Discussion · 15 min · optionalSpeaker note: The creative-writing case is the trap. RAG doesn't help much there since there's no 'ground truth' document to retrieve from. Use the debrief to surface that.
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Quiz · 15 min · optionalSpeaker note: The 'both' scenarios are deliberately included. Make sure to call out which ones need both RAG and tool calling and why.
Slide 16
Slide 17
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Chapter 4. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Chapter 4. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Icebreaker · 15 min · optionalSpeaker note: Pick a bug that's a genuine state/props misunderstanding (e.g. stale closure, missing dependency), not a typo. It should test reading comprehension, not eagle eyes.
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Discussion · 15 min · optionalSpeaker note: Use two real v0/Lovable outputs from the same prompt session if you have them. The inconsistency is more convincing when it's real.
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Quiz · 15 min · optionalSpeaker note: This pairs the quiz with a live keyboard-only test if possible. The gap between visual guess and actual keyboard test is the lesson.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Chapter 5. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Chapter 5. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Icebreaker · 15 min · optionalSpeaker note: Pick app ideas with a genuinely defensible answer either way to get real debate, not a unanimous vote.
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Design Challenge · 15 min · optionalSpeaker note: The 'who can edit what' question is the bridge into the authorization content coming next. Steer the debrief there.
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Opinion · 15 min · optionalSpeaker note: There's no clean right answer. The point is surfacing that 'the AI should have caught this' isn't currently a safe assumption.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Chapter 6. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Chapter 6.
Slide 23
Speaker note: Final key-term definitions for Chapter 6. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Icebreaker · 15 min · optionalSpeaker note: Light and fast. Sets up the 'read the docs' lesson with real, relatable frustration.
Slide 5
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Design Challenge · 15 min · optionalSpeaker note: Push groups toward concrete answers: 'check the signature' and 'store the event ID and skip duplicates' are the two things to listen for.
Slide 15
Slide 16
Slide 17
Slide 18
Slide 19
Slide 20
Slide 21
Slide 22
Quiz · 15 min · optionalSpeaker note: This is a fast pattern-recognition drill for the 'never trust the client' instinct that recurs through Chapters 7, 13, and 14.
Slide 23
Slide 24
Slide 25
Slide 26
Speaker note: Definitions for the key terms in Chapter 7. Walk these before the recap; students can self-check against the chapter glossary.
Slide 27
Speaker note: Continued key-term definitions for Chapter 7. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Icebreaker · 15 min · optionalSpeaker note: The point isn't a perfectly clean taxonomy. It's getting students to articulate WHY something does or doesn't have the plan, act, observe loop.
Slide 5
Slide 6
Slide 7
Slide 8
Slide 9
Discussion · 15 min · optionalSpeaker note: The invoice-processing task is deliberately ambiguous (could be either, depending on how standardized the invoices are). Use it to surface the actual decision criteria.
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Slide 18
Slide 19
Discussion · 15 min · optionalSpeaker note: Push students to name what could break (inbox, calendar, GitHub, payments) instead of collecting channels for novelty.
Slide 20
Slide 21
Slide 22
Quiz · 15 min · optionalSpeaker note: There usually isn't one perfect answer. Score for sound reasoning about stakes and reversibility, and use disagreement as the teaching moment.
Slide 23
Slide 24
Slide 25
Speaker note: This pair is the whole capability story. Demos live in the first number; students' ambitions usually live in the second.
Slide 26
Slide 27
Discussion · 15 min · optionalSpeaker note: Push toward the hierarchy: most groups discover the portal has an API, which is the pixels-last lesson landing on its own.
Slide 28
Slide 29
Slide 30
Speaker note: Potential / alternate Chapter 8 lab. Students who can't install still complete the design path for full credit.
Slide 31
Slide 32
Speaker note: Definitions for the key terms in Chapter 8. Walk these before the recap; students can self-check against the chapter glossary.
Slide 33
Speaker note: Continued key-term definitions for Chapter 8. Pair with the recap chips so terms land before the chapter closes.
Slide 34
Speaker note: Personal-agent terms for Section 8.6. Emphasize that these are scaffolding names, not a new model family.
Slide 35
Speaker note: The three terms added with Section 8.7, in the same format as the preceding definition slides.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Icebreaker · 15 min · optionalSpeaker note: Have a backup prompt ready in case the room can't easily trigger one. Newer models hallucinate less on common requests, so obscure/edge-case prompts work best.
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Slide 11
Slide 12
Discussion · 15 min · optionalSpeaker note: Pull a real example of SEO slop from a recipe site or similar, and pair it against a genuinely well-written short passage. The 'almost fooled you' cases are the most instructive.
Slide 13
Slide 14
Slide 15
Slide 16
Opinion · 15 min · optionalSpeaker note: Push toward the realization that 'passed tests' is necessary but not sufficient. Tests only check what someone thought to test.
Slide 17
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Chapter 9. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Chapter 9. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Icebreaker · 15 min · optionalSpeaker note: Pick a function with a genuinely non-obvious edge case (off-by-one, empty list, type coercion) so predictions actually diverge.
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: Pick a function where the missing edge case genuinely breaks it, so the live test run pays off with a real failure.
Slide 18
Slide 19
Slide 20
Slide 21
Slide 22
Quiz · 15 min · optionalSpeaker note: The point is showing that gut-feel ranking and rubric-based scoring sometimes disagree. That's exactly why benchmarks exist.
Slide 23
Slide 24
Slide 25
Slide 26
Speaker note: Definitions for the key terms in Chapter 10. Walk these before the recap; students can self-check against the chapter glossary.
Slide 27
Speaker note: Continued key-term definitions for Chapter 10. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Icebreaker · 15 min · optionalSpeaker note: The goal is building the instinct to recognize vision-model opportunities in everyday tasks, not building anything yet.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Discussion · 15 min · optionalSpeaker note: This is meant to surface genuine disagreement and nuance. There's no single correct answer, but everyone should leave able to articulate the disclosure principle from the chapter.
Slide 14
Slide 15
Slide 16
Slide 17
Quiz · 15 min · optionalSpeaker note: Use real pipeline examples (voice assistant, document-to-summary, photo-to-caption-to-speech) so the 'weak seam' reasoning transfers to real builds.
Slide 18
Speaker note: The fetch-vs-see distinction is the payoff of this slide. An agent can quote a page perfectly and still not know the header is broken.
Slide 19
Slide 20
Speaker note: Connect this back to 11.5: decomposition is the chapter's pipeline thinking, performed by the agent on its own input. Same seams, same skepticism.
Slide 21
Demo · 15 min · optionalSpeaker note: Pick the bug in advance so the contrast is guaranteed. The markdown answer will be confidently blind; the screenshot answer will spot it.
Slide 22
Slide 23
Slide 24
Slide 25
Speaker note: Definitions for the key terms in Chapter 11. Walk these before the recap; students can self-check against the chapter glossary.
Slide 26
Speaker note: Continued key-term definitions for Chapter 11. Pair with the recap chips so terms land before the chapter closes.
Slide 27
Speaker note: The three terms added with Section 11.6, defined in the same format as the previous two slides.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Icebreaker · 15 min · optionalSpeaker note: Some of these answers may genuinely seed final-project ideas later in the course. Worth noting that out loud.
Slide 9
Slide 10
Slide 11
Slide 12
Design Challenge · 15 min · optionalSpeaker note: Watch for groups that list 'non-goals' that are actually still implied requirements. Push them to be honest about what they're really cutting.
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Quiz · 15 min · optionalSpeaker note: Use real or plausible product ideas students would recognize. The close calls are where the defensibility discussion actually happens.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Chapter 12. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Chapter 12. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Icebreaker · 15 min · optionalSpeaker note: Use a fake/expired key for the demo, obviously. But make the snippet realistic enough that it takes a genuine second look to find.
Slide 11
Slide 12
Slide 13
Slide 14
Discussion · 15 min · optionalSpeaker note: There's no single right answer per app, but the reasoning process (find the narrowest pipe) is the actual lesson.
Slide 15
Slide 16
Slide 17
Slide 18
Quiz · 15 min · optionalSpeaker note: Use the actual current per-token pricing for a real model so the exercise produces a genuinely useful number, not just an abstraction.
Slide 19
Slide 20
Slide 21
Slide 22
Speaker note: Definitions for the key terms in Chapter 13. Walk these before the recap; students can self-check against the chapter glossary.
Slide 23
Speaker note: Continued key-term definitions for Chapter 13. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Icebreaker · 15 min · optionalSpeaker note: Provider policies live on a 'data usage' or developer-docs page, not the general consumer privacy policy. Point students there directly if they get stuck.
Slide 5
Slide 6
Slide 7
Slide 8
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Design Challenge · 15 min · optionalSpeaker note: Steer toward 'treat email content as data, not instructions' plus a guardrail limiting what the agent can do regardless of what it's told.
Slide 14
Slide 15
Opinion · 15 min · optionalSpeaker note: There's no clean answer here. The point is practicing the 'is this proportionate' judgment rather than reaching a verdict.
Slide 16
Slide 17
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Chapter 14. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Chapter 14. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Opinion · 15 min · optionalSpeaker note: 15-min opener. Use AlgoArena Classroom to collect answers, then cluster them live. Most reasons will be 'it removed real pain,' which previews A.1.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Design Challenge · 15 min · optionalSpeaker note: 15-min build-and-pitch. Pairs swap and give one piece of 'is this real demand?' feedback. Great warm-up for a final-project pitch.
Slide 9
Slide 10
Slide 11
Slide 12
Speaker note: Definitions for the key terms in Optional A. Walk these before the recap; students can self-check against the chapter glossary.
Slide 13
Speaker note: The remaining terms, in the same format as the preceding definition slide.
Slide 1
Slide 2
Slide 3
Icebreaker · 15 min · optionalSpeaker note: 15-min opener via AlgoArena Classroom poll. The point lands when the class is split or wrong. Surface-level detection is already unreliable, which motivates the whole chapter.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Quiz · 15 min · optionalSpeaker note: 15-min AlgoArena Classroom quiz. Use it to check the transition from B.1 to B.2: students should be able to say *why* slop is produced, not just that it exists.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min design sprint in pairs. The 'how could it be gamed?' prompt is the real lesson. Every trust signal becomes a new target for slop.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix B. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix B. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Icebreaker · 15 min · optionalSpeaker note: 15-min opener. Capture choices in AlgoArena Classroom. The recurring answer 'it removed a specific friction' previews the C.1 design lens.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Discussion · 15 min · optionalSpeaker note: 15-min group discussion. The tension (a guardrail can feel like friction) is the lesson: removing the 'annoying' confirm step is how products start shipping mistakes.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min solo teardown, then 3 volunteers present. Encourage live edge-case probing on a laptop. The architecture inferences are the crowd-pleaser.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix C. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix C. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Opinion · 15 min · optionalSpeaker note: 15-min opener. Collect both stories in AlgoArena Classroom. The contrast is the productive-struggle paradox in the students' own words before you name it.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Discussion · 15 min · optionalSpeaker note: 15-min pair exercise. The teaching prompt almost always asks for explanation or critique, not answers. That's the takeaway to surface in debrief.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min build-and-break. The 'try to break it' step is the real test. Most first attempts are still paste-solvable until they require judgment on a specific, novel artifact.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix D. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix D. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Icebreaker · 15 min · optionalSpeaker note: 15-min opener via AlgoArena Classroom. Most stories are about absent rationale, not hard algorithms. That previews E.2's reverse-engineering of intent.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Quiz · 15 min · optionalSpeaker note: 15-min AlgoArena Classroom quiz. Grade the *process* answer ('add a log here', 'check this value'). Jumping to a fix is the anti-pattern to call out.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min pair exercise. The disagreement is the point. Make students articulate the risk of deletion, not just their gut feeling.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix E. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix E. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Icebreaker · 15 min · optionalSpeaker note: 15-min opener. Collect in AlgoArena Classroom and group by domain (legal, product, security). You'll reuse these as the live case studies through the chapter.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Discussion · 15 min · optionalSpeaker note: 15-min discussion. Push past 'I'd just notice'. The lawyers thought so too. The goal is to name a concrete verification step.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min synthesis. Connect each case to a specific core chapter's practice. This is the slide that turns 'scary stories' into a checklist.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix F. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix F. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Opinion · 15 min · optionalSpeaker note: 15-min opener via AlgoArena Classroom. The convincing-but-wrong examples preview G.3's 'presentation outrunning substance' directly.
Slide 4
Slide 5
Slide 6
Slide 7
Discussion · 15 min · optionalSpeaker note: 15-min group exercise. The ranking is the lesson. Most auto-generated charts answer questions nobody needed answered.
Slide 8
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min lab. If laptops are available, run it live. The speed at which 'findings' appear is itself the cautionary demo.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix G. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix G. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Opinion · 15 min · optionalSpeaker note: 15-min opener via AlgoArena Classroom. The 'health records / source code / legal docs' answers preview H.2's privacy-as-hard-requirement framing.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Quiz · 15 min · optionalSpeaker note: 15-min AlgoArena Classroom quiz. Reward the reasoning, not the label. The same scenario can go either way depending on which constraint dominates.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min spec exercise. If possible, demo `ollama run` on a small model live so the download-and-go simplicity is concrete.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix H. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix H. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Opinion · 15 min · optionalSpeaker note: 15-min opener via AlgoArena Classroom. Surface that prompting solves more than beginners expect. That's the core message of I.1.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Quiz · 15 min · optionalSpeaker note: 15-min AlgoArena Classroom quiz. The trap answers are tasks better solved by better prompting or retrieval. Name why fine-tuning would be overkill.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min planning exercise. The held-out discipline is the whole grade here. If the test set leaked into training, the plan proves nothing.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix I. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix I. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Icebreaker · 15 min · optionalSpeaker note: 15-min opener via AlgoArena Classroom. The 'getting users is scarier' consensus sets up J.2: building is now the easy half.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Discussion · 15 min · optionalSpeaker note: 15-min discussion. The class should arrive at 'non-coding skills decide it'. That's the bridge into distribution and bootstrapping.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min exercise. Reject any 'go viral' answers. Force a concrete, repeatable channel. This doubles as final-project distribution planning.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix J. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix J. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Opinion · 15 min · optionalSpeaker note: 15-min opener via AlgoArena Classroom. Save the predictions. Revisit them after K.1 so students see whole-job 'replacement' is rarer than the task-level shift.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Discussion · 15 min · optionalSpeaker note: 15-min group exercise. The insight to surface: almost no job is all one bucket, which is why 'AI replaces jobs' is too coarse a claim.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min synthesis. Tie the 'human owns' column back to the verification/judgment skills the course trains. That's the durable-career payoff.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix K. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix K. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Opinion · 15 min · optionalSpeaker note: 15-min opener via AlgoArena Classroom. Record the 'yes, unprecedented' share now and revisit at the end. The count usually softens, which is the chapter's point.
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Quiz · 15 min · optionalSpeaker note: 15-min AlgoArena Classroom quiz. Blind the dates if you can. Students often can't tell a 1985 claim from a 2024 one, which is exactly the lesson.
Slide 9
Slide 10
Slide 11
Slide 12
Slide 13
Slide 14
Slide 15
Slide 16
Slide 17
Design Challenge · 15 min · optionalSpeaker note: 15-min exercise. The structural match (not the topic) is the grade. Students should spot the same overshoot shape across decades.
Slide 18
Slide 19
Slide 20
Slide 21
Speaker note: Definitions for the key terms in Appendix L. Walk these before the recap; students can self-check against the chapter glossary.
Slide 22
Speaker note: Continued key-term definitions for Appendix L. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
Slide 8
Slide 9
Speaker note: Definitions for the key terms in Appendix M. Walk these before the recap; students can self-check against the chapter glossary.
Slide 10
Speaker note: Continued key-term definitions for Appendix M. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Speaker note: Definitions for the key terms in Appendix N. Walk these before the recap; students can self-check against the chapter glossary.
Slide 7
Speaker note: Continued key-term definitions for Appendix N. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
Speaker note: Definitions for the key terms in Appendix O. Walk these before the recap; students can self-check against the chapter glossary.
Slide 8
Speaker note: Continued key-term definitions for Appendix O. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Speaker note: Frame the chapter as practical setup plus one habit. By the end of the session, every student should have a working key and a checkpointed repo.
Slide 2
Speaker note: The point to land: students who checkpoint attempt bigger things, because failure becomes an inconvenience instead of a loss.
Slide 3
Slide 4
Speaker note: The demo that lands: have the agent recover 'the version of X from before yesterday's change' live. Students who have lost work to an eager model feel this one immediately.
Slide 5
Demo · 15 min · optionalSpeaker note: Run this live if projecting. The reveal is how much larger the diff is than the agent's summary suggested.
Slide 6
Slide 7
Slide 8
Quiz · 15 min · optionalSpeaker note: Ask which file goes to GitHub before showing the answer. Getting this backwards is the single most common error in the chapter.
Slide 9
Slide 10
Discussion · 15 min · optionalSpeaker note: Drive to the answer: rotate first. Push protection matches secret patterns, not filenames, so DB_PASSWORD=hunter2 sails through.
Slide 11
Slide 12
Slide 13
Speaker note: Definitions for the key terms in Appendix P. Walk these before the recap; students can self-check against the chapter glossary.
Slide 14
Speaker note: Continued key-term definitions for Appendix P. Pair with the recap chips so terms land before the chapter closes.
Slide 1
Speaker note: Set up the chapter as the money chapter. Chapter 4 treated tokens as a ceiling, Chapter 1 treated tiers as a quality trade-off, and this one puts a price on both.
Slide 2
Slide 3
Speaker note: This is the load-bearing idea of the chapter. Don't move on until the room believes it.
Slide 4
Slide 5
Quiz · 15 min · optionalSpeaker note: Almost everyone underestimates, and the reason is always the same: they counted what they typed, not what was re-sent. Let a few students say their numbers out loud before revealing the class spread.
Slide 6
Slide 7
Slide 8
Slide 9
Opinion · 15 min · optionalSpeaker note: Ask the room to name the most expensive thing they have done to save money on tokens. Someone will describe re-rolling a mid-tier model four times, and that story teaches the section better than the table does.
Slide 10
Design Challenge · 15 min · optionalSpeaker note: The overspend matters more than the estimate. Push pairs that only optimize downward to name where cheapness would cost them an afternoon.
Slide 11
Slide 12
Speaker note: Definitions for the key terms in Appendix Q. Walk these before the recap; students can self-check against the chapter glossary.
Slide 13
Speaker note: Continued key-term definitions for Appendix Q. Pair with the recap chips so terms land before the chapter closes.
Five multiple-choice questions drawn from the core chapters. Answers and explanations follow each question.
Q1. In the vibecoding loop (intent, generation, verification, refinement), which two stages belong to the human?
Answer: B. The model owns first-draft generation. You own intent (saying clearly what you want) and verification (deciding whether you actually got it). Refinement is the conversation that closes the gap.
Q2. Why does a human's verification role become MORE important as AI tools gain agency, not less?
Answer: C. A chat assistant that's wrong costs you a bad copy-paste. An agentic IDE that's wrong can change ten files and run a destructive command before you've finished reading its plan, so an unverified mistake can do far more damage.
Q3. Which tool category lets the model read multiple files, propose multi-file diffs, AND execute commands inside your project?
Answer: D. Executing commands inside your project (running tests, starting a dev server, reading the error it just produced) is the defining capability of an AI-native IDE. Chat assistants can't touch your filesystem by default, and app builders sit a layer above the editor.
Q4. A model confidently uses a function from a library version that doesn't exist. What's this an example of?
Answer: A. Hallucination is when a model produces plausible-sounding but false output, such as a nonexistent API. It's especially likely near a model's training cutoff or for fast-moving libraries, which is why you verify against real docs.
Q5. Why does the SAME underlying model sometimes feel much smarter in one tool than another?
Answer: B. The wrapper around a model controls which files, errors, and instructions reach it. A disappointing result often means the tool didn't show the model the right context, not that the model couldn't do the task, which is why curating context is a core skill.