[HN Gopher] Launch HN: Golpo (YC S25) - AI-generated explainer v...
___________________________________________________________________
Launch HN: Golpo (YC S25) - AI-generated explainer videos
Hey HN! We're Shraman and Shreyas Kar, building Golpo
(https://video.golpoai.com), an AI generator for whiteboard-style
explainer videos, capable of creating videos from any document or
prompt. We've always made videos to communicate any concept and
felt like it was the clearest way to communicate. But making good
videos was time-consuming and tedious. It required planning,
scripting, recording, editing, syncing voice with visuals. Even a
2-minute video could take hours. AI video tools are impressive at
generating cinematic scenes and flashy content, but struggle to
explain a product demo, walk through a complex workflow, or teach a
technical topic. People still spend hours making explainer videos
manually because existing AI tools aren't built for learning or
clarity. Our solution is Golpo. Our video generation engine
generates time-aligned graphics with spoken narration that are good
for onboarding, training, product walkthroughs, and education. It's
fast, scalable, and built from the ground up to help people
understand complex ideas through simple storytelling. Here's a
demo: https://www.youtube.com/watch?v=C_LGM0dEyDA#t=7. Golpo is
built specifically for use cases involving explaining, learning,
and onboarding. In our (obviously biased!) opinion, it feels
authentic and engaging in a way no other AI video generator does.
Golpo can generate videos in over 190 languages. After it generates
a video, you can fully customize its animations by just describing
the changes you want to see in each motion graphic it generates in
natural language. It was challenging to get this to work!
Initially, we used a code-generation approach with Manim, where we
fine-tuned a language model to emit Python animation scripts
directly from the input text. While promising for small examples,
this quickly became brittle, and the generated code usually
contained broken imports, unsupported transforms, and poor timing
alignment between narration and visuals. Debugging and regenerating
these scripts was often slower than creating them manually. We
also explored training a custom diffusion-based video model, but
found it impractical for our needs. Diffusion could produce high-
fidelity cinematic scenes, but generating coherent sequences beyond
about 30 seconds was unreliable without complex stitching, making
edits required regenerating large portions of the video, and
visuals frequently drifted from the instructional intent,
especially for abstract or technical topics. Also, we did not have
the compute to scale this. Existing state-of-the-art systems like
Sora and Veo 3 face similar limitations: they are optimized for
cinematic storytelling, not step-by-step educational content, and
they lack both the deterministic control needed for time-aligned
narration and the scalability for 5-10 minute explainers. In the
end, we took a different path of training a reinforcement learning
agent to "draw" whiteboard strokes, step-by-step, optimized for
clear, human-like explanations. This worked well because the action
space was simple and the environment was not overly complex,
allowing the agent to learn efficient, precise, and consistent
drawing behaviors. Here are some sample videos that Golpo
generated: https://www.youtube.com/watch?v=33xNoWHYZGA (Whiteboard
Gym - the tech behind Golpo itself)
https://www.youtube.com/watch?v=w_ZwKhptUqI (How do RNNs work?)
https://www.youtube.com/watch?v=RxFKo-2sWCM (function pointers in
C) https://golpo-podcast-inputs.s3.us-east-2.amazonaws.com/file...
(basic intro to Godel's theorem) You can try Golpo here:
https://video.golpoai.com, and we will set you up with 2 credits.
We'd love your feedback, especially on what feels off, what you'd
want to control, and how you might use it. Comments welcome!
Author : skar01
Score : 108 points
Date : 2025-08-13 17:11 UTC (1 days ago)
(HTM) web link (video.golpoai.com)
(TXT) w3m dump (video.golpoai.com)
| typs wrote:
| If that demo video is how it actually works, this is a pretty
| amazing technical feat. I'm definitely going to try this out.
|
| Edit: I've used. It's amazing. I'm going to be using this a lot.
| skar01 wrote:
| Thank you!!
| mclau157 wrote:
| I have used AI in the past to learn a topic but by creating a GUI
| with input sliders and output that I can see how things change
| when I change parameters, this could work here where people can
| basically ask "what if x happens" and see the result which also
| makes them feel in control of the learning
| skar01 wrote:
| Thank you!!
| skar01 wrote:
| Hey also, if you want to suggest a video, we could try generating
| one and reply here with a link! Just tell us what you want the
| video to be about!!
| cube2222 wrote:
| Hey, kudos for the product / demo on the website - it managed
| to keep me engaged to watch it till the end.
|
| I'm mostly curious how it fairs with more complex topics and
| doing actually informative (rather than just "plain
| background") illustrations.
|
| Like a video explaining transformer attention in LLMs, to stay
| on the AI topic?
| skar01 wrote:
| Yeah so it actually does pretty well. Here are some sample
| videos:
|
| https://www.youtube.com/watch?v=33xNoWHYZGA&t=1s
|
| https://www.youtube.com/watch?v=w_ZwKhptUqI
| andhuman wrote:
| Could you do a video about latent heat?
| metalliqaz wrote:
| So... if I had the enterprise accounts for various LLM services,
| could I dupe this company with a basic upload page and a nice big
| prompt?
| Wolf_Larsen wrote:
| Its not that simple, but it would be straight forward to
| duplicate the outputs of this with a simple LLM + ffmpeg
| workflow. They did mention a custom model on the landing page,
| and if they've trained one then you would be spending much more
| money on each output than they are. Because without a fine-
| tuned model there would be a lot of inference done for QA and
| refinement of each prompt | clip | frame .
| MarcelOlsz wrote:
| "Custom model" usually translates to "deployed an OSS model
| and tweaked a few things" like 99% of the time.
| Lienetic wrote:
| I'm curious - do you feel differently about some of these
| coding and coding-adjacent tools out there like Cursor and
| Lovable?
| metalliqaz wrote:
| no, not really. I think they are massively over-valued but in
| the tech world... what else is new? I view those tools as
| mostly a convenience. They are integrating things into nice
| easy packages to use. That's the value.
|
| With this... eh. Most people don't need to make more than one
| or two explainer videos, so are they going to take on a new
| monthly fee for that? And then there are power users who do
| it all the time, but almost surely have their own workflow
| put together that is customized to exactly what they want.
|
| At any point, one of the big players could introduce this as
| a feature for their main product.
| poly2it wrote:
| The creator tier ($99.99/mo) lists "15 seconds" as a perk. Does
| this mean the maximum video length is 15 seconds?
| bangaladore wrote:
| Given that the next tier up is "Create longer/more detailed
| video (up to 4 min long)", I'd guess you are right.
|
| Seems like this is pretty useless unless you pay 200$ per
| month. Which may be a reasonable number for the clearly
| commercial / enterprise use case, but I'm just not certain what
| you can do wtih the lower tiers.
| skar02 wrote:
| One of the founders here! No it's not. The max video length is
| up to 2 min, which is also the case in any non-free tier. We
| just include a 15-second option for that tier (because people
| it need for things like FB ads)
| poly2it wrote:
| Maybe clarify it a bit. Eg. "Short 15 second option".
| BugsJustFindMe wrote:
| In the post you talk about 5-10 minute explainers.
|
| What does one do if they want to make a 5-10 minute explainer
| if the maximum length is 2 minutes?
| metalliqaz wrote:
| My suggestion would be to re-think the demo videos. I have only
| watched most of the way into the "function pointers in C"
| example. If I didn't already know C well, I would not be able to
| follow that. The technical diagrams don't stay on the screen long
| enough for new learners to process the information. These videos
| probably look fantastic to the person who wrote the document it
| summarizes, but to a newbie the information is fleeting and hard
| to follow. The machine doesn't understand that the screen
| shouldn't be completely wiped all the time while it follows the
| narrative. Some visuals should be static for paragraphs, or stay
| visible while detail marked up around it. For a true master of
| the art, see 3blue1brown.
| bangaladore wrote:
| > For a true master of the art, see 3blue1brown.
|
| I agree. Rather than (what I assume is) E2E text -> video/audio
| output, it seems like training a model on how to utilize the
| community fork of manim which 3blue1brown uses for videos would
| produce a better result.
|
| [1] https://github.com/ManimCommunity/manim/
| albumen wrote:
| Manim is awesome and I'd love to see that, but it doesn't
| easily offer the "hand-drawn whiteboard" look they've got
| currently.
| WasimBhai wrote:
| I have 2 credits but it won't let me generate a video. Founders,
| if you are around, you may want to debug.
| skar02 wrote:
| Huh, that's odd. Could you DM me your email?
| skar01 wrote:
| Or just email us at founders@golpoai.com
| delbronski wrote:
| Wow, I was skeptical at first, but the result was pretty awesome!
|
| Congrats! Cool product.
|
| Feedback: I tried making a product explainer video for a tree
| planting rover I'm working on. The rover looked different in
| every scene. I can imagine this kind of consistency may be more
| difficult to get right. Maybe if I had uploaded a photo of how
| the rover looks it may have helped. In one scene the rover looks
| like an actual rover, in the other it looks like a humanoid
| robot.
|
| But still, super impressed!
| skar01 wrote:
| Thanks! We are working on the consistency.
| KaoruAoiShiho wrote:
| Did NotebookLM just come out with this? Very tough to compete
| with google.
| empressplay wrote:
| Can confirm, it creates slides though, not whiteboard
| animations. Although the slides are in color and have graphs,
| clipart, etc. (but they are static and the whiteboard drawing
| is cooler!)
|
| It created an 8 minute video explaining my Logo-based coding
| language using 50 sources and it was free.
|
| https://www.youtube.com/watch?v=HZW75burwQc
| skar01 wrote:
| We have color as well and support graphs and clipart
| adi4213 wrote:
| This is neat but I wasn't able to get it to work (server
| overloaded is what the browser app said) I'd also recommend
| registering a custom domain in Supabase so the Google SSO shows
| the golpo domain - which is a small, but professional-signaling
| affordance
| skar01 wrote:
| We will soon! Wanted to get the model working first! Could you
| try again
| ishita159 wrote:
| Planning to add links as input anytime soon?
|
| I would love to add a link to my product docs, upload some images
| and have it generate an onboarding video of the platform.
| skar02 wrote:
| Yes, very soon. We already support this via API and will add to
| our platform too!
| skar01 wrote:
| Our API is currently available to our enterprise customers!
| reactordev wrote:
| This is actually pretty amazing. Not only does it work, it's
| good. At least from the demo videos. YMMV.
|
| What I always wanted to do was to teach what I know but I lack
| the time commitment to get it out. This might be a way...
| skar01 wrote:
| Thank you so much!
| CalRobert wrote:
| So it eats concepts and makes videos?
|
| One is reminded of smbc
|
| https://www.seekpng.com/png/detail/213-2132749_gulpo-decal-f...
| skar02 wrote:
| Haha! The name actually comes from the word story in Bengali.
| ceroxylon wrote:
| The generated graphic in the linked demo for "Training materials
| that captivate" is a sketch of someone looking forlorn while
| holding a piece of paper. Is there a way to do in-line edits to
| the generated result to polish out things like this?
| skar01 wrote:
| We are working on that. There will ultimately be a storyboard
| feature where you can edit frame by frame!
| nextworddev wrote:
| Has anyone tried prompting VEO to create these videos
| skar02 wrote:
| We have! Veo I believe, can't do more than 8-second videos, and
| when prompted they aren't very coherent in our experience.
| nextworddev wrote:
| oh had no idea. will try your product
| OG_BME wrote:
| I created a video on the free tier, the shareable link didn't
| work (404), I upgraded to be able to download it, and it seems to
| have disappeared? It says "Still generating" in my Library.
|
| The video UUID starts with "f5fbd6c7", hopefully that's
| sufficient to identify me!
| skar02 wrote:
| Sorry about that! I found your video. Should I link it here or
| DM it to you (can you do DM in Hacker News?) ? You could also
| email me at shreyas2@stanford.edu, and I can send it there
| dang wrote:
| (No DMs on HN, at least not yet)
| OG_BME wrote:
| Just emailed you! Thanks.
| Lienetic wrote:
| This is really interesting, definitely going to give it a try!
| Seems fun but are you seeing people actually needing to make lots
| of videos like this? What's your vision - how does this become
| really big?
| drawnwren wrote:
| I'm sure someone else has mentioned this but your video on the
| main page correctly has GRPO the first time it's introduced but
| then every time you mention it after that -- you've swapped it to
| GPRO.
| tk90 wrote:
| Pretty cool, especially the voice and background music - feels
| just right.
|
| I asked it about pointers in Rust. The transcript and images were
| great, very approachable!
|
| "Do not let your computer sleep" -> is this using GPU on my
| machine or something?
| skar01 wrote:
| No! We just had that because we had not built the library
| feature yet, and just forgot to remove it. Now you can access
| through there!!
| subhro wrote:
| From one Kar to another, duurdaant glp Congratulations.
| skar02 wrote:
| Thanks!
| albumen wrote:
| Love it. The tone is just right. A couple of suggestions:
|
| Have you tried a "filled line" approach, rather than "outlined"
| strokes? Might feel more like individual marker strokes.
|
| I made a demo video on the free tier and it did a great job
| explaining acoustic delay lines in an accessible fashion, after
| feeding it a catalog PDF with an overview of the historical
| artefact and photography of an example unit. Unfortunately the
| service invented its own idea of what the artefact looked like.
| Could you offer a storyboard view and let users erase the
| incorrect parts and sketch their own shapes? Or split the drawing
| up into logical elements and the user could redraw them as
| needed, which would then be reused where that element is used in
| other frames?
| skar01 wrote:
| Thank you!! We are actually currently working on the
| storyboarding feature!!
| BoorishBears wrote:
| Very cool: what output format is the model producing?
|
| Straight vector paths?
| dtran wrote:
| Love this idea! The Whiteboard Gym explainer video seemed really
| text-heavy (although I did learn enough to guess that that's
| because text likely beat drawing/adding an image for these
| abstract concepts for the GRPO agent). I found Shraman's personal
| story video much more engaging!
| https://x.com/ShramanKar/status/1955404430943326239
|
| Signed up and waiting on a video :)
|
| Edit: here's a 58s explainer video for the concept of body
| doubling: https://video.golpoai.com/share/448557cc-
| cf06-4cad-9fb2-f56b...
| addandsubtract wrote:
| The body doubling concept is something I've noticed myself, but
| never knew there was a term for it. TIL :)
| ActVen wrote:
| Popup window with "Load Failed" after it had some progress on the
| bar past 40% or so. Shows up in the library, but won't play. I
| just deleted it for now.
| skar01 wrote:
| Could you try again?
| ActVen wrote:
| Just tried on Chrome instead of safari and it worked this
| time. Thanks and congrats on the launch!
| skar01 wrote:
| Thank you!
| meistertigran wrote:
| Can you share the paper mentioned in the demo video?
| trenchpilgrim wrote:
| I threw the user docs for my open source project in there and it
| was... surprisingly not terrible!
|
| Note: Your paywall for downloading the video is easily bypassed
| by Inspect Element :)
|
| My main concern for you is that y'all will get Sherlocked by
| OpenAI/Anthropic/Google.
| mkagenius wrote:
| Not only the giants. They will face significant threat from
| open source too[1]. But they just need to carve their own user
| base and be profitable in that space.
|
| 1. For example, I have built http://gitpodcast.com which can be
| run for free. Can also be self hosted using free tier of gemini
| and azure speech.
| ayaros wrote:
| In the Khan academy videos I remember watching, an instructor
| would actually write on a tablet; you'd see each letter get hand-
| written one by one in order. Is there no way to get it to do
| that? What the AI is doing instead is building-up the strokes of
| every character on the line of text all at once, which looks
| completely unnatural. The awkwardness is compounded by the fact
| that the letters are outlined, so it takes even more steps to
| create them.
|
| In addition, the line-art style of the illustrations looks like
| that same cartoonish-AI-slop style I see everywhere now. I just
| can't take it seriously.
|
| If this tool is widely deployed it's just going to get used to
| spread more misinformation. I'm sure it will be great for bad
| actors and spammers to have yet another tool in their toolbox to
| spread whatever weird content or messages they want. But for the
| rest of us, that means search engines and YouTube and other
| places will be filled with a million AI-generated half-baked
| inferior copies of Khan Academy. It's already hard enough to find
| good educational resources online if you don't know where to
| look, and this will only make the problem worse.
|
| You'll just have to forgive me if I'm not really excited about
| this tool.
|
| ...also the name is a bit weird. It reminds me of "Gulpo, the
| fish who eats concepts" from that classic SMBC cartoon.
| (https://www.smbc-comics.com/comic/2010-12-15)
| mandeepj wrote:
| Congrats on the launch!
|
| If I may ask - how do you generate your audio?
| raylad wrote:
| Feedback on the text: I find the way that the text generates
| randomly across the line very distracting because I (and I think
| most people) read from left to right. Having letters appear
| randomly is much more difficult to follow.
|
| Are there options to have the text appear differently?
| dfee wrote:
| From the video
|
| > The Al needs to figure out not just what to draw, but
| precisely when to draw it
|
| ;)
| sdotdev wrote:
| I'll try the 1 free generation soon, but the way the text appears
| randomly in that landing page demo video is really weird. I keep
| loosing track of where I'm reading too as the audio sometimes is
| not perfectly synced. The sync is not that bad however, but it
| could be better.
| ks2048 wrote:
| I made it 8 seconds into the "function pointers in C" video and
| immediately stopped. It went too fast to read the code examples
| and diagrams. (second "slide" appears for 1 second.. and what is
| that array it is showing?) If you go back and look at the code (a
| three line swap function) - it's messed up. No opening bracket
| and where is the closing bracket? It is "swaps first and last",
| but hard-coded to only length 3 arrays?
|
| I'm sure AI could help make good animations like this, but this
| looks like slop.
| personjerry wrote:
| I feel like this is another case of throwing AI in a non-AI-
| required problem. Khan Academy itself just hired people to make
| its videos at a very reasonable wage. Why would you need to add
| AI into the equation? If you wanted to, you could build a
| platform of basic video / whiteboard content creators at a very
| reasonable price point.
| wordpad wrote:
| You can't have arbitrary content with a human in the workflow.
| personjerry wrote:
| You can absolutely hire a human to make arbitrary content
| atleastoptimal wrote:
| someone needs to do something about the purple darkmode rounded
| corner tailwind style that has infected all LLMs now.
|
| cool product though!
| UltraSane wrote:
| Impressive. Reminds me of Google NotebookLLMs AI generated
| podcasts of PDFs.
| android521 wrote:
| Do you have a developer api that empowers developers to create
| explainer videos?
| giorgioz wrote:
| I love the concept but the implementation in the demo seem not
| good enough to me. I think the black and white demo is quite
| ugly... 1) Explainer videos are not in black and white 2) the
| images are not drawn live usually. 3) text being drawn on the go
| is just a fake animation. In reality most explainer videos show
| short meaningful sentence appearing all at once so the user has
| more time for reading me.
|
| Keep up refining the generated demo! Best of luck
| fxwin wrote:
| I'm also not the biggest fan of the white-on-black style, but
| there is definitely precedent (at least in science-youtube-
| space) for explainer videos "drawn live" [1-4]
|
| [1] https://www.youtube.com/@Aleph0
|
| [2] https://www.youtube.com/@MinutePhysics
|
| [3] https://www.youtube.com/@12tone
|
| [4] https://www.youtube.com/@SimplilearnOfficial
| whitepaint wrote:
| I've tried it and it is really cool. Well done and good luck.
| torlok wrote:
| Going by the example videos, this is nothing like I'd expect a
| whiteboard video to look like. It fills the slides in
| erratically, even text. No human does that. It's distracting more
| than anything. If a human teacher wants to show cause-and-effect,
| they'll draw the cause, then an arrow, then the effect to
| emphasize what they're saying. Your videos resemble printing more
| than drawing.
| grues-dinner wrote:
| It seems really strange that you wouldn't farm this kind of
| thing out to a non-AI function that animates the text properly
| into the space given using parameters that the AI generated. I
| mean it's impressive that it does work at all, let alone as
| well as it does, but are we also going to get an AI to do all
| the video encoding as well?
| achempion wrote:
| Where I can find what is a credit? It says 150 credits for a
| Growth plan but doesn't explain how many credits are needed for a
| single video
|
| p.s. the pricing section is unreadable under the 840px width
| snowfield wrote:
| I want to pay 20usd just to troll my friends with explainer
| videos on why they're shit at video games :D
| ludicrousdispla wrote:
| that seems like excellent product market fit as the AI
| generated explainer videos won't even need to be correct, and
| the more incorrect they are the better the troll
| ing33k wrote:
| it created this video for an app I am working on.
| https://video.golpoai.com/share/8de80271-1109-48e4-ac52-9265...
| qwertytyyuu wrote:
| The way text is appears is so weird, is like rendering a by
| plotting each letter aschronously. I wonder how it compares to
| auto generated power point presentations. I suspect it might be
| worse
| futhey wrote:
| That worked. Really well!
|
| But, white on black is really ugly. Even black on white or a
| simple inversion would be an improvement.
|
| I think it could benefit from the ability to pause and see the
| transcript, and make edits before the video is generated.
| Terretta wrote:
| Chalkboards are white on black. You basically have chalkboards
| or whiteboards to draw from. Not sure either is landing in the
| Louvre. Both have their aesthetic uses. I'd imagine chalkboards
| for academic topics, whiteboards for business, but different
| ages and cultures will feel differently.
___________________________________________________________________
(page generated 2025-08-14 23:02 UTC)