{
  "video": {
    "id": "s0hFne6EeOI",
    "title": "Build the Right Thing: Product Engineering (Part 2) — Kent C. Dodds, EpicProduct.engineer",
    "duration": 3129,
    "upload_date": null,
    "channel": "AI Engineer",
    "source": "AI Engineer",
    "event_format": ""
  },
  "analysis": {
    "video_id": "s0hFne6EeOI",
    "title": "Build the Right Thing: Product Engineering (Part 2) — Kent C. Dodds, EpicProduct.engineer",
    "one_liner": "When AI agents level the implementation playing field, building the right thing becomes the differentiator. Use jobs-to-be-done to turn a feature request into the real problem behind it, and the Kano model to decide how much engineering effort each change deserves.",
    "summary": "In the second half of his product-engineering workshop, Kent C. Dodds uses an example conference app, \"Just Get In the Workshop\", to show how a product engineer should dig past a feature request (facial recognition) to the job to be done: getting attendees into a workshop quickly without missing the first 20 minutes. He applies Clay Christensen's jobs theory, including the functional/social/emotional dimensions and the big hire vs. little hire, and uses it to explain why React won. He then uses a food-delivery demo of the 1984 Kano model to show that basics, performance needs and delighters each call for different engineering investment, and that features only move from delighter toward basic need over time.",
    "key_points": [
      "Keep asking why until you reach the job to be done. Map out the 'problem tree': each solution creates offshoot problems, and if you had known how expensive a downstream problem would be, you might have gone a different direction higher up.",
      "Job statement template: 'When [situation], help me [make progress] so I can [desired outcome], without [cost].' The worked example was 'When there's a workshop going on at 10:10, help me get into the workshop quickly so that I can learn about product engineering without missing the first 20 minutes.' It has nothing to do with facial recognition, and alternatives such as parallel lines or NFC badges fall out of it.",
      "Break the job into functional (does it do the thing: state, workflow, integrations such as a face database, encryption, venue capacity), social (who is involved: the door staff, people in line, the speaker) and emotional (for example being creeped out that the conference holds your facial data). Each dimension maps to technical decisions.",
      "React won first for functional reasons (faster than Backbone, Ember and AngularJS, a simpler state model, good composition and encapsulation), then for social reasons around 2015–2016 (bosses, teams and jobs required it), and network effects made later functional advantages in SolidJS or Vue irrelevant. Open-source libraries are products too.",
      "Jobs theory separates the 'big hire' (the purchase) from the 'little hire' (every use, like every bite of a burger), so success is defined by repeated progress, not the initial sale.",
      "Kano model demo on a food-delivery app: unreliable order confirmation emails are a basic need, ETA accuracy is a performance need, a surprise discount is a delighter, and shared cart/group ordering is a delighter moving toward performance. Live driver GPS was a delighter in 2015 and is now a basic need. Features only move from delighter to performance to basic, or drop out entirely if nobody cares.",
      "Kano also has indifferent features (dead weight to remove) and reverse features (things users hate, such as auto-posting check-ins to social media). Engineers build those shady features, so they should push back.",
      "Incident response: Dodds told his agent 'production is down, figure it out' when a bot attacked his DigitalOcean-hosted personal site, and it diagnosed and fixed the problem. That is fine for low-stakes systems, but at PayPal scale (millions of dollars lost quickly) you have to understand the code and architecture yourself."
    ],
    "takeaways": [
      "Never turn feature requests directly into implementation, and especially not by piping them into an agent automatically. If a task arrives without a job to be done attached, go back to the PM before designing the system.",
      "Validate a request against other users with Mom Test questions such as 'When was the last time you experienced this pain and what did you do instead?' rather than 'Is this a problem for you?'. The user is often directionally right but not always right.",
      "Set up a feedback channel (for example a Slack channel that pulls in feedback from high-value customers, Reddit and X) so you can see the paper cuts. Treat apparent user error as a system design problem.",
      "Size engineering effort by Kano category: basics need reliability, completeness and ownership; performance needs measurable quality targets you instrument; delighters are experiments that can live in a separate system until their shape is clear. Get the foundations right before investing in delighters.",
      "Revisit prioritisation regularly, because competitors' delighters become your performance needs and then your basic needs. Challenge indifferent features that cost maintenance time, such as two weeks migrating something nobody uses."
    ],
    "topics": [
      "product engineering",
      "jobs to be done",
      "kano model",
      "prioritization",
      "user research",
      "ai agents",
      "product judgment"
    ],
    "tools": [
      "React",
      "Backbone",
      "Ember",
      "AngularJS",
      "Angular 2",
      "SolidJS",
      "Vue",
      "Bun",
      "Node",
      "PayPal",
      "DigitalOcean",
      "Cloudflare Pages",
      "Slack",
      "Reddit",
      "X",
      "Discord",
      "YouTube",
      "Super Duper Burgers",
      "Competing Against Luck (Clay Christensen)",
      "The Mom Test",
      "EpicProduct.engineer"
    ],
    "quotes": [
      {
        "text": "So, you do not just turn feature requests into implementation.",
        "at": "07:25",
        "url": "https://www.youtube.com/watch?v=s0hFne6EeOI&t=445s"
      },
      {
        "text": "what is the toy that you give your pet agent to be happy like that that's all that really matters at this point",
        "at": "12:40",
        "url": "https://www.youtube.com/watch?v=s0hFne6EeOI&t=760s"
      },
      {
        "text": "If there's no toilet paper in my hotel, that's a bad experience. But if I have a hundred rolls of toilet paper, it doesn't increase my experience with the hotel.",
        "at": "44:12",
        "url": "https://www.youtube.com/watch?v=s0hFne6EeOI&t=2652s"
      },
      {
        "text": "When AI agents level the implementation playing field, the differentiator becomes building the right thing.",
        "at": "51:01",
        "url": "https://www.youtube.com/watch?v=s0hFne6EeOI&t=3061s"
      }
    ],
    "words": 11478
  },
  "summary_url": "/#s0hFne6EeOI",
  "transcript": {
    "html": "/transcripts/s0hFne6EeOI.html",
    "txt": "/transcripts/s0hFne6EeOI.txt",
    "vtt": "/transcripts/s0hFne6EeOI.vtt"
  }
}