Show HN: I made a website that converts YT videos into step-by-step guides

Overall reception & use cases

• Many commenters find the idea highly useful, especially for:
  • DIY, repair, and technical tutorials (solar, car repair, NAS setup, coding).
  • Recipes and other step-by-step tasks where video fluff is frustrating.
  • People who strongly prefer text over video for learning and reference.
• Some consider it something they’ve “been looking for for ages” and see it as one of the more practical LLM applications.

Feature suggestions & UX

• Frequent requests:
  • Per-step timestamps and deep links back into the video.
  • Use YouTube chapters where available.
  • Checklist mode for following along while doing the task.
  • Screenshot/thumbnail per step, possibly à la iFixit-style guides.
  • Loading indicators and better handling of multi-second processing delays.
  • Browser extension and/or a searchable database of “stepified” videos.
• Suggestions to curate or reorder “Recent” videos and moderate NSFW or prank submissions.

Technical implementation & reliability

• Current approach: pull YouTube transcripts via API and feed them to an LLM.
• For videos without captions, it currently relies on the description, which causes odd or low-quality guides.
• Multiple people recommend using Whisper or similar models as a fallback for missing captions and possibly integrating SponsorBlock to skip ads and self-promo.
• Site frequently hits Heroku/internal errors; suggestions include:
  • Asynchronous processing with a job queue.
  • Notifications when processing is done.
  • Possibly using Cloudflare or scaling hosting.

Accuracy, limitations, and edge cases

• Works well for some tutorials and recipes, but:
  • Can include “like/subscribe/comment” lines as steps unless explicitly filtered.
  • Produces gibberish or hallucinations on non-tutorial content.
  • Some users report that for complex mechanical tasks the output is not reliable enough to follow.
• Debate over whether different LLMs (e.g., non-OpenAI) might do better at summarization.

Creator rights, copyright, and ethics

• Major subthread on ethics:
  • Concerns about generating derivative textual guides without creator consent, especially when:
    • Pages are permanent.
    • There’s no deletion or contact mechanism.
    • The site asserts “all rights reserved” in the footer.
  • Worries about:
    • SEO competition with creators’ own content.
    • Reduced incentive to watch the original video and loss of income.
    • Inaccurate summaries reflecting poorly on the original video.
• Counterarguments:
  • The site embeds the YouTube player; views and ad revenue still go to the creator.
  • Embedding and summarizing publicly available videos is compared to news sites and blogs that embed and describe videos.
  • Creators can disable embedding on YouTube; critics respond that summaries are still generated even then.
• Some suggest opt-in, clearer attribution (title, description, channel link at the top), and/or revenue sharing or tools specifically for creators.

Business model, openness, and sustainability

• The developer is currently paying significant API costs out-of-pocket, with mention of burning a large fraction of a paycheck.
• Discussion about whether to open-source the tool vs. turning it into a business; no clear decision.
• Some envision paid “pro” features (checklists, extensions, advanced navigation) as a possible path.