Humans& Raises $480M Seed Round to Build AI with Social Intelligence

The era of one-on-one AI assistants is evolving into something far more complex: collective intelligence. Humans&, a startup founded by a "dream team" of alumni from Anthropic, OpenAI, and Google DeepMind, just announced a massive $480 million seed round. According to TechCrunch, this capital will fund the development of a "central nervous system" for human-AI collaboration.

Humans& AI Collaboration Seed Round 2026: Beyond Individual Chatbots

At just three months old, Humans& has already sparked a bidding war among investors. The company isn't just building another chatbot; it's designing a foundation model for social intelligence. CEO Eric Zelikman, formerly of xAI, argues that while current models are smart, workflows aren't. Humans& aims to close this gap by creating AI that can coordinate people with competing priorities and track long-running decisions.

The Technical Frontier: Long-Horizon and Multi-Agent RL

To achieve this, the team is focusing on Long-horizon Reinforcement Learning (RL) and Multi-agent RL. Unlike typical LLMs optimized for immediate user satisfaction, these models are trained to plan, act, and revise over time. Co-founder Yuchen He noted that the model needs to remember things about itself and the users to function as "connective tissue" across organizations.

The stakes are high. Humans& is entering a field crowded by established giants. Anthropic is pushing Claude Cowork, while Google embeds Gemini directly into Workspace. Even smaller players like Granola—which recently raised $43 million at a $250 million valuation—are vying for the collaboration layer.