← Back to AI Insights
Gemini Executive Synthesis

Scrimba Explain (demoed as HN.watch), an LLM-powered platform for generating HTML-based explainer videos from text.

Technical Positioning
A fast, cheap, and easily editable alternative to pixel-based video generation, enabling video explanations for virtually any text content (e.g., pull requests, documentation, articles, course drafts).
SaaS Insight & Market Implications
Scrimba Explain addresses a significant bottleneck in content creation: the cost and time associated with video production. By generating HTML-based videos at ~$0.04 and in seconds, it fundamentally alters the economics of video content. This enables pervasive video integration across enterprise workflows, from internal documentation and pull request explanations to rapid course drafting. The "cents and seconds" value proposition unlocks use cases previously impractical due to traditional video production overhead. While visual fidelity may be a trade-off, the speed and cost benefits for informational content are compelling. The proprietary stack (Imba, OP, Q) demonstrates a full-stack approach to optimize LLM interaction and content generation, indicating a strategic move to control the entire value chain for this new video format. This positions Scrimba to capture a substantial share of the instructional and informational video market.
Proprietary Technical Taxonomy
HTML-based video format LLM explainer videos diffusion models pixel-based videos cost per video ~$0.04 AI-assisted editing Imba

Raw Developer Origin & Technical Request

Source Icon Hacker News Sep 29, 2026
Show HN: HN.watch – Videos of all Hacker News posts

Hi HN, I’m Per, founder of Scrimba (YC S20). We’ve spent the last decade teaching people how to code with an HTML-based video format. We’ve now plugged an LLM into it, so that people can create explainer videos about anything. It’s called “Scrimba Explain”.To demo this technology for Hacker News, we built HN.watch. It’s like HN, but with explainer videos instead of articles. We create them on-the-fly the first time someone clicks on a link.While there are obvious visual drawbacks of using HTML instead of diffusion models, there are three big benefits: - Speed: Much faster to generate than pixel-based videos (just a few seconds from click to playback) - Cost: Our cost per video is ~$0.04. (Excluding image generation, which some videos utilize. Quickly blows up the cost) - Easy editing: the above benefits also make AI-assisted editing cheap & fastOur hypothesis is that if video creation goes from “dollars and minutes” to “cents and seconds”, a bunch of new use cases will be unlocked. Here are some we see already: - A video explanation of every single Pull Request (we do this internally) - Give every page in your internal/extrernal docs a video - Turn a complex article into a video in ~4 seconds (via our Chrome extension) - Course creators can quickly draft lessons before recording the real thing - People also create a lot of personal stuff stories for their kids, wedding invitations, birthdays, etcThe stack is based on an open-source programming language (Imba) created by our CTO, Sindre Aarsæther. It compiles to JavaScript, so it interoperates fully with the npm + node ecosystem. You can learn more here: imba.io/We’ve also built our own sync engine (OP), and a context management system for agents (Q). We feared this would make the LLMs struggle when writing code for us, as neither is in their training data (there’s very little Imba in there too). However, we’ve been pleasantly surprised to see that LLMs actually are really good at our stack. This is probably because the stack is extremely dense. Imba is compact, and so is OP, where a single declaration sets storage, sync, permissions, UI, and what the AI sees. This means there’s no translations between frontend, API, db and JSON where the model can get confused and get things wrong.Simply said, instead of using React.js, Express, Supabase, and LangChain, we built it all from scratch. Definitely suffering from the “not invented here” syndrome, lol! As for the models, we use Gemini, GPTs, Inworld, ElevenLabs, and a few others.If you want to try it out, just take your pick: - The Web UI (scrimba.com/explain) - MCP (add it to your coding agent) - ChatGPT Plugin - Chrome ExtensionYou can find a link to all of the above in our docs: docs.scrimba.com/explain/introduct... finally, a real pixel-based video of the tool:

love to hear your feedback and if anyone has ideas for other use cases.PS: I expect quite a bit of pushback from HN for this launch, given how fan of text the HN crowd is. This kind of tool is not for everyone. But there are a lot of people today who prefer videos over text, especially in the younger generations.

Developer Debate & Comments

spoon404 • Sep 29, 2026
I really dislike this but have to say it is impressive.To test it I tried "Who Killed Paulina Borsook's Career?" The summary is correct I did not hear any errors. https://www.wired.com/story/paulina-borsook-profile/ https://hn.watch/?item=49866903It did do what I expected from such a system: - It removed the fun parts of the article and the way it is written. - Stated some interpretations as fact without naming the source Article '"That gunshot prematurely aged my brain,” she said' Summary 'The impact prematurely aged her brain'One surprise is did not like: The sharpest concrete thing was a .45 caliber bullet. Feels like some sort of attempt to add original humor and I do not like it.This is the worst kind of article for this kind of summary and I feel like it really shows the most and least impressive parts of the system.
totallygeeky • Sep 28, 2026
Clicked on the demo "What caused the fall of the Roman Empire?" and immediately saw the same inadequacies I've seen in every other system like this: any attempt at visualizing a point ends up being absolute nonsense. Graphics and charts that confuse more than help, or even contradict what is being said.It ends up being a hurdle I don't think these services can surpass, given these systems understand nothing, and have no spatial reasoning. Any of these systems that claim to help explain things visually end up being a huge liability for the learner and I would never recommend them to someone.
vbernat • Sep 28, 2026
This looks a good idea to let people choose their preferred medium. I am not into video either, but I turned one of my latest blog article into a video using Claude, since I thought it would be “easy.” I asked it to build the video by screencasting a browser presenting the interactive diagrams of the article and add some titles. It build some kind of recording app for me. It still took me many hours, but I think I would be faster if I redid it. I would pay to have it done faster, but I want to keep control of the text, not have a summary.The article: https://vincent.bernat.ch/en/blog/2026-spanning-tree-videoThe tool built by Claude to make the video: https://github.com/vincentbernat/vincent.bernat.ch/tree/2808...
russellbeattie • Sep 28, 2026
I just tried it out - it's pretty cool. The postscript was actually noted in the generated video about this topic, which I find amusing.I have to say, if HN were to add an AI generated paragraph summary about each link at the top of each comments page, it would probably go a long way to improving the discussions about a lot of topics. The number of people who comment based on the title of a link alone is pretty high, and I'd bet the vast majority of commenters only skim the linked pages anyways. Might as well get everyone on the same page with a quick overview.
customguy • Sep 28, 2026
> And finally, a real pixel-based video of the tool: https://www.youtube.com/watch?v=k6rbHmBxSEs"a compressor uses three main organs" @ 2:97I don't dislike that it's video (because I know people who won't read can't be "made to read" by there not being "nice enough explainer videos"), but this is the kind of stuff that bothers me in both video and text, and greatly. I don't want to rant about it here because it's not specific to your product at all, it's not even specific to LLM, the internet and especially youtube is full of stuff that neither speakers/authors nor audience seem to ever parse. And if you're just downstream of big models you probably can't really influence that. But since you are NIH enjoyers, maybe you could train your own at some point? I don't want to consume such videos, but I do want those who do to have nicer ones, because that'd be good for me, too.
fishtoaster • Sep 28, 2026
Because we all contain multitudes, I can simultaneously accept that:1. I hate everything about this because I vastly prefer text over video for the same content, especially AI generated video2. There are a lot of people for whom video is their preferred medium and so this will be valuable to them.It's a technically cool project and your cost-per-video is impressively low. Best of luck!
dverlaeckt80 • Sep 28, 2026
This is actually quite impressive from an engineer viewpoint. I just have the feeling that the videos quickly become very monotonous and rather boring due to the monotonous AI voices. If somehow you could bring dynamic variation in these videos that would be fantastic.
Vaslo • Sep 28, 2026
This is great - many of us who are technical but not working in hardcore tech probably scroll by quickly on stuff we have never heard of. Getting a quick synopsis like this may drive more traffic and understanding.Thanks!
scosman • Sep 28, 2026
These AI explainers are taking over. I tried getting this working back in May. The models were okay but not quite there and it was a ton of effort. Opus 5.5 seems like the tipping point.I made an OSS framework for these for when you want to go beyond one-shoting it: https://github.com/scosman/videowright- Voiceovers: aligns animations to the voiceover, can generate voiceover with elevenlabs, or will transcribe and timestamp a real voiceover- can reorder scenes both in code, and using ffmpeg for audio.- interactive controls during authoring, can ask for micro edits or re-builds- MP4 export/encoder- Generates a video from a prompt (obvs)
harvey9 • Sep 28, 2026
Please, nobody click on the link to this hn item within hn.watch as it would be even more dangerous than typing 'google' into Google

Frequently Asked Questions

Market intelligence mapped to Scrimba Explain (demoed as HN.watch), an LLM-powered platform for generating HTML-based explainer videos from text..

What problem does Scrimba Explain (demoed as HN.watch), an LLM-powered platform for generating HTML-based explainer videos from text. solve?
Based on our AI analysis of the original developer request, its primary technical positioning is: A fast, cheap, and easily editable alternative to pixel-based video generation, enabling video explanations for virtually any text content (e.g., pull requests, documentation, articles, course drafts).
Are engineers actively discussing Scrimba Explain (demoed as HN.watch), an LLM-powered platform for generating HTML-based explainer videos from text.?
Yes, we have tracked 97 direct responses and active debates regarding this specific topic originating from Hacker News.
Which technical concepts are associated with Scrimba Explain (demoed as HN.watch), an LLM-powered platform for generating HTML-based explainer videos from text.?
Our proprietary extraction maps Scrimba Explain (demoed as HN.watch), an LLM-powered platform for generating HTML-based explainer videos from text. to adjacent architectural concepts including HTML-based video format, LLM, explainer videos, diffusion models.

Engagement Signals

226
Upvotes
97
Comments

Cross-Market Term Frequency

Counts stored text matches for foundational terms like LLM and Gemini in analysed public discussions. These counts do not measure adoption.

Stored text matches, not adoption or market-share measurements.

  • LLM: 219 records
  • Gemini: 43 records
  • JavaScript: 15 records
  • Supabase: 12 records
  • Express: 11 records
  • LangChain: 5 records
  • HTML-based video format: 1 records
  • explainer videos: 1 records
  • diffusion models: 1 records
  • pixel-based videos: 1 records
  • cost per video ~$0.04: 1 records
  • AI-assisted editing: 1 records
  • Imba: 1 records
  • npm + node ecosystem: 1 records
  • sync engine (OP): 1 records
  • context management system for agents (Q): 1 records
  • React.js: 1 records
  • GPTs: 1 records
  • Inworld: 1 records
  • ElevenLabs: 1 records
  • ChatGPT Plugin: 1 records
  • Chrome Extension: 1 records