Launching TwelveLabs on Product Hunt again - Lessons learned
TwelveLabs just introduced Pegasus 1.5, their most significant leap in generative video AI, transforming video into a queryable, structured data asset.
They're launching today on Product Hunt.
Fun fact: This is their 2nd official launch there.
Previously, the video AI company launched Marengo 3.0 in December 2025. [1] They ranked among the Top Products and got featured in Product Hunt's newsletters. This new launch keeps the momentum going.
Once again, I had the opportunity to collaborate with the team on this new launch and wanted to share with you some lessons learned and how we approached this launch.
TL,DR: Keep it simple, and keep launching.
Lessons learned launching on Product Hunt
Planning a new launch starts with the required inputs:
Name of the launch: 40 characters max. We kept it simple: Pegasus 1.5 by TwelveLabs.
Tagline: 60 characters max. The tagline might be the most important part of a launch, [2] so we made multiple iterations to find a balance, highlighting the key features while keeping it relatable.
Description: 500 characters max. We made it shorter, in 246 characters.
Launch tags: up to 3 tags.
Thumbnail: Product Hunt supports GIF files so we were able to add a nice animated image (only plays on hover).
Image gallery: Product Hunt recommends adding 3 or more images. Above all, we wanted to show the product. We framed product screenshots and added a headline and subheadline for context.
Pro tip: You can find the required inputs on GitHub. [3]
Product Hunt is about consistency
Above all, the #1 lesson learned is that launching on Product Hunt definitely isn't a one-time opportunity.
Keep it simple, and keep launching. That's it!
TwelveLabs is launching today Pegasus 1.5, their most significant leap in generative video AI, transforming video into a queryable, structured data asset.
Replies
The "keep launching" point is underrated. Most people treat it like a one shot event when it's clearly not.
exactly - launch, and keep launching. @Product Hunt pays off in the long term.
Great read, loved all the launch lessons. I’ve been curious about how Twelve Labs compares to Google’s NotebookLM, especially when it comes to structuring information from videos & docs. Could you share your thoughts on the main differences between the two? I’m still figuring out which one fits better for structured understanding workflows.
TwelveLabs
@alfredo_b_grabowski Yes! Pegasus 1.5 is built around generating structured, time-based metadata across entire videos — it's designed to turn raw video into queryable, production-ready structured outputs. TwelveLabs has other models like Marengo 3.0, which can create video embeddings, and use text, images, video clips, or audio to find specific content. NotebookLM, by contrast, is a document comprehension and synthesis tool to explore and interrogate text-based sources; thus, analysis on documents would probably best be done on NotebookLM.
If you had any amount of video that needed structured understanding, TwelveLabs would be best at producing the type of custom structured output you'd expect.
Document + Video analysis is also something that TwelveLabs is working on -- can I ask what questions you are interested in getting structured understanding on?
@alfredo_b_grabowski Do you feel one is better for deep research workflows?
@alfredo_b_grabowski How often do you actually work with video data?
@alfredo_b_grabowski Do you care about real time querying or batch results?
@alfredo_b_grabowski Is your workflow more research driven or production focused?
Really interesting shift — turning video into structured, queryable data feels powerful.
From a UX side, the value is clear, but I wonder how intuitive this is for non-technical users.
Feels like the real challenge is not just the tech, but making this interaction simple enough to actually use daily.
Curious — what was the hardest part of designing the UX for this?
TwelveLabs
Hi @ivan_anisimov4 - The "Time-based Metadata Extraction" is quite developer-centric — JSON schema in, structured JSON out — as you mentioned. And that's intentional: it is best suited for developers and teams building on top of video. The structured output is yours to parse, store, and query — it's how you keep your video data organized and useful downstream.
But the design challenge was exactly what you're pointing at: non-technical users still need to understand the value without having to stare at bracket soup. We leaned into examples heavily, built a visual schema builder side-by-side with the raw JSON so the structure is legible even if you've never touched a schema before, and shipped templates around the most common use cases :) Really lowering that technical hurdle for users to try and see its value.
minimalist phone: creating folders
So TL;DR of it is making it more visual and shorter! :D
exactly - keep it simple, and keep launching
I loved the idea that you just need to keep launching and not be afraid. Many teams obsess too much over the perfect first time and end up burning out. The concept of turning video into structured data is very much in demand right now.
TwelveLabs
@milenap Thanks Milena! Appreciate the support!
@milenap spot on! just launch
Nice breakdown. Iterating on the tagline and keeping the launch simple is underrated. Consistency really seems to be the edge here.
@nithikanb 100% - the tagline might actually be the most important part of a launch. see this related thread.
Scade.pro
Thank you for the detailed instructions!
thank you, Maria! anything that struck you?
Scade.pro
@fmerian "launching on Product Hunt definitely isn't a one-time opportunity" - 100% true
do you usually prioritize clarity over creativity, or try to push both equally?