Sunain's () bet: Architectures will change. Data richness compounds. 2026 is the year of multimodal AI.

x.com/i/article/2011…

all the labs that wanna scale the embodied ai know that the solution isn’t more data but data from the right environments. if you go on to hugging face today, there is sooo much high quality, super structured footages from labs and warehouses. its all pick and place, same motions

scaling human data by orders of magnitude is the only path toward general intelligence

i love helping researchers out in every way i can. if you're a researcher working on robotics/vlas/world models and need some data to play with, dm me.

just heard back from a researcher at a major robotics foundation model company: “...multi-cam, multi-task, multi-environment home data moves performance way more than people think, one of the highest-leverage signals for generalization yet.”

Sunain gives over 15000 everyday people an opportunity to be part of this growing AI economy. These people take part in training the frontier models across all modalities which makes sure that every person understands how these models work and their advantages to the humaity.

We built Sunain on a simple belief: the future of AI should be trained by real people, in real environments, across real cultures, not just synthetic data and lab setups. Excited to see it come to life with Protege !

Yesterday we announced our exciting new partnership with - check out some clips of their authentic, unscripted human behavior data! 👥 10,000+ contributors... 🏁 across 30+ countries... 📣 50+ languages record conversations, play games, and capture real-world

Humans are the only scalable resource for AGI