humans&

Founded by Eric Zelikman in 2026

Build frontier models that understand human relationships and assist collaborative teamwork

Pricing model

Paid

Rating

★ Mixed Reviews

Time to value

Requires Expertise

You can use humans& to access next-generation foundation models designed around human behavior and team collaboration. The lab develops systems using multi-agent reinforcement learning, persistent memory, and deep user understanding to connect groups and support complex shared work. Instead of treating synthetic agents purely as isolated assistants, it focuses on strengthening relationships across organizations and communities. Researchers and developers can study behavioral training methods, interact with cooperative autonomous agents, and integrate models built to handle long-horizon tasks while preserving human context and trust.

What humans& does

Train models using multi-agent reinforcement learningSimulate human collaboration dynamics with autonomous agentsMaintain persistent memory across long context windowsEvaluate model alignment in social interaction benchmarksCoordinate complex group tasks across multiple systemsDeploy cooperative agent workflows for workplace teamsMulti-agent reinforcement learning for group coordinationLong-horizon planning and reasoning capabilitiesPersistent memory systems for extended human contextDeep user understanding models focused on interpersonal trustBehavioral alignment training across collaborative environmentsContributions to open source AI research and academic development