/Tech3h ago

Medal Builds Largest Video-Action Dataset for World Model Training

6151117.6K

Original post

on their @latentspacepod we covered how @pimdewitte accidentally made the PERFECT world model data collection business by collecting the world's largest dataset of trainable (video,action) pairs.

turning the attention economy into the attention industry.

congrats Pim!!

link in reply

Pim de Witte@PimDeWitte

Today we're announcing our $320M Series A at a $2.3B valuation, led by @khoslaventures , with @generalcatalyst, @JeffBezos , @ericschmidt, @NicoRosberg, alongside leading researchers from frontier labs and academia.

12:48 PM · Jun 25, 2026 · 5.4K Views

Sentiment

Users back Medal's $320M Series A at $2.3B valuation because its real-world 'vision in, actions out' data loop and organic flywheel create distinctive advantages over static scraping.

Pos

100.0%

Neg

0.0%

2 comments with sentiment.

Cluster Engagement

Digg Deeper

No Digg Deeper questions have been answered for this story yet.

YOUTUBEVia

#225

Posts from X

Most Activity

VIEWS2.2KBOOKMARKS1

swyx 🔜 @aiDotEngineer@swyx

@latentspacepod @PimDeWitte https://www.youtube.com/watch?v=A2P3Q3LCoLw

swyx 🔜 @aiDotEngineer@swyx

on their @latentspacepod we covered how @pimdewitte accidentally made the PERFECT world model data collection business by collecting the world's largest dataset of trainable (video,action) pairs.

turning the attention economy into the attention industry.

congrats Pim!!

link in reply

3h2.2K01

Artem@at56_

@swyx @latentspacepod @PimDeWitte the flywheel nobody drew up on purpose is always the best one

3h8

Tech News@tech_summaries

the "vision in, actions out" loop is exactly why this data is different. most labs are scraping static video; having the actual input labels (mouse, keys, triggers) makes it a world model, not just a video predictor.

turning that pipeline into a $2.3B valuation makes total sense.

3h2