Humans& Targets AI‑Driven Coordination with New Foundation Model

Humans&, a startup founded by veterans of Anthropic, Meta, OpenAI, xAI, and Google DeepMind, is building a foundation model focused on social intelligence and team coordination. The company raised a large seed round to develop a “central nervous system” that can help people collaborate, make group decisions, and interact with AI in a more conversational way. The model will be trained with long‑horizon and multi‑agent reinforcement learning to remember users, understand motivations, and act as connective tissue across organizations. While the product is still in development, the team aims to own the collaboration layer rather than plug into existing tools.

Background and Vision

Humans& was created by a group of former researchers from leading AI labs including Anthropic, Meta, OpenAI, xAI, and Google DeepMind. The founders see the next major frontier for artificial intelligence as the ability to coordinate people with competing priorities, track long‑running decisions, and keep teams aligned over time. Their ambition is to move beyond the current wave of chat‑oriented models and build a new foundation model architecture designed for social intelligence, not just information retrieval or code generation.

Funding and Strategy

The startup secured a substantial seed round, allowing it to pursue the development of what the founders describe as a “central nervous system” for the human‑plus‑AI economy. Rather than creating a model that merely plugs into existing collaboration platforms, Humans& intends to own the coordination layer itself, acting as the connective tissue across any organization, from large enterprises to families.

Product Direction

While no product has been released yet, the team envisions a solution that could replace or augment multi‑user contexts such as communication platforms (e.g., Slack) and collaborative document tools (e.g., Google Docs, Notion). The focus is on helping people work together and communicate more effectively, both with each other and with AI tools. The model will be trained to ask questions in a friendly, colleague‑like manner, aiming to understand the value of each query rather than simply optimizing for immediate user approval.

Technical Approach

Humans& plans to train its model using long‑horizon reinforcement learning, which encourages the system to plan, act, revise, and follow through over extended periods. Multi‑agent reinforcement learning will also be employed so the model can operate in environments where multiple AIs and humans interact simultaneously. This training regime is intended to give the model memory of itself and its users, improving its ability to understand motivations, skills, and needs across a group.

Market Context and Competition

The space of AI‑enhanced collaboration tools is heating up, with other startups and major AI firms launching features that integrate AI into workflow platforms. Companies such as Anthropic, Google, and OpenAI are already embedding collaborative AI capabilities into their products. However, none of these players appear to be building a model centered on social intelligence from the ground up, which could give Humans& a unique advantage or make it an attractive acquisition target.

Risks and Outlook

Developing a new foundation model is capital‑intensive and requires massive compute resources, putting Humans& in direct competition with established AI giants for both funding and infrastructure. The startup has already turned down acquisition interest, emphasizing its belief that it can become a generational company that fundamentally changes how humans interact with AI.