Latent Space AI HumansAnd Raises $480M Seed Round to Build AI for Human Collaboration
Jan 26, 2026
A deep dive into a startup raising $480M to build AI that coordinates group decision making. Discussion covers designing models for social intelligence and memory. The conversation examines multi-agent training strategies and the practical risks of scaling, hiring, and competing with big players.
AI Snips
Chapters
Transcript
Episode notes
Models For Social Intelligence
- HumansAnd aims to build models optimized for social intelligence and group coordination rather than solo question-answering.
- The startup positions its model as a "central nervous system" for human+AI collaboration across teams and households.
Coordination Is The Real Bottleneck
- Many current AI bottlenecks are coordination and fragmented workflows, not just model intelligence.
- Workers feel threatened by efficiency promises, creating adoption and trust challenges for AI tools.
Seed Money Before A Product
- Despite a huge seed raise, HumansAnd didn't yet have a public product and kept its first vision intentionally vague.
- The team prefers to co-evolve model behavior with interface design rather than plug into existing collaboration tools.