Co-hunt with @sentry_co to share TwelveLabs, an AI platform focused entirely on deep video understanding – going beyond just frame analysis or basic metadata.
The name itself looks familiar in the current AI audio/voice landscape... 😉 But it's actually something about video AI, aims to let you interact with video content in powerful new ways.
Powered by their multimodal models (Marengo for search/embeddings, Pegasus for text generation), you can use natural language to semantically search for moments within large video libraries, generate detailed summaries or analyses, and create rich embeddings for tasks like classification or recommendations.
It handles large-scale libraries and offers a free tier and playground to get started. Seems like a robust platform for anyone working seriously with video data.
Really enjoyed the demo playground and the use case! In a previous life I used to run an agency and lead shoots - and my biggest nightmare was logging footage and parsing MONTHS after we were done with campaigns. Would've loved to have this at that point!
Curious if ya'll end up including an edit tool for shortlisted footage at any point?
I hate having to source footage for my content and nothing shows up because the videos are spread over 20 different sites. This sounds super handy.
honestly feels like Tony Stark would use to analyze CCTV footage. Might mess around and find my lost sunglasses from 2020.
Wow man! It's really impressive. You're bringing video understanding to the next level and I'm sure many content creators will love this. Wish you all the best!
