Imagine trying to get a dozen colleagues to agree on a company logo. The endless emails, conflicting opinions, and scheduling nightmares – it’s the kind of messy human coordination that current AI assistants can’t handle. But a new startup called Humans& thinks this coordination gap represents the next major frontier in artificial intelligence, and they’ve just raised $480 million to prove it.
Founded by alumni from Anthropic, Meta, OpenAI, xAI, and Google DeepMind, Humans& aims to build what they call a “central nervous system” for the human-plus-AI economy. Their pitch? Moving beyond AI as just a question-answering tool to creating models designed specifically for social intelligence and team coordination.
Beyond Chatbots to Collaboration
“It feels like we’re ending the first paradigm of scaling, where question-answering models were trained to be very smart at particular verticals,” says Andi Peng, co-founder and former Anthropic employee. “Now we’re entering what we believe to be the second wave of adoption where the average consumer or user is trying to figure out what to do with all these things.”
The timing couldn’t be more critical. Companies are transitioning from simple chat interfaces to AI agents, but while models have become competent, workflows haven’t kept pace. The coordination challenge – managing people with competing priorities, tracking long-running decisions, and keeping teams aligned – remains largely unaddressed.
Eric Zelikman, CEO and former xAI researcher, explains their vision: “We are building a product and a model that is centered on communication and collaboration. The focus is on getting the product to help people work together and communicate more effectively – both with each other and with AI tools.”
The Technical Challenge: Rethinking AI Training
To achieve this, Humans& is rethinking how AI models are trained. Yuchen He, co-founder and former OpenAI researcher, reveals they’re using long-horizon and multi-agent reinforcement learning (RL). Long-horizon RL trains models to plan, act, revise, and follow through over time rather than just generating one-off answers. Multi-agent RL prepares models for environments where multiple AIs and humans interact.
“The model needs to remember things about itself, about you, and the better its memory, the better its user understanding,” He explains. This approach represents a significant shift from current models optimized primarily for immediate user satisfaction and answer accuracy.
The Broader AI Landscape: Physical and Agentic Expansion
Humans& isn’t operating in a vacuum. The AI landscape is expanding in multiple directions simultaneously. According to a Financial Times report, physical AI – the convergence of AI and robotics – is revolutionizing industries beyond traditional manufacturing. Over 4.7 million industrial robots were in operation in 2024, with annual installation growth exceeding 500,000 units.
Stephan Schlauss, global head of manufacturing at Siemens, notes that “AI-enabled robots that pick and place different parts and materials in our assembly lines reduce automation costs by 90 percent.” This physical AI expansion complements the coordination focus of companies like Humans&, creating a comprehensive AI ecosystem that spans both digital and physical domains.
The Safety Challenge: Racing Ahead of Protocols
As AI adoption accelerates, safety concerns are mounting. A Deloitte report surveying over 3,200 business leaders across 24 countries reveals that businesses are deploying AI agents faster than safety protocols can keep up. Currently, 23% of companies use AI agents moderately, projected to jump to 74% in two years, yet only 21% have robust safety mechanisms.
The report warns: “Given the technology’s rapid adoption trajectory, this could be a significant limitation. As agentic AI scales from pilots to production deployments, establishing robust governance should be essential to capturing value while managing risk.” This safety gap becomes particularly relevant for coordination-focused AI like what Humans& proposes, where models would have access to sensitive team dynamics and organizational information.
Competition and Market Dynamics
Humans& faces formidable competition. They’re not just competing with collaboration tools like Slack and Notion – they’re taking on AI giants. Anthropic is developing Claude Cowork for work-style collaboration, Google has embedded Gemini into Workspace, and OpenAI is pitching developers on multi-agent orchestration.
Yet Humans& has one potential advantage: none of the major players seem poised to rewrite their models based on social intelligence. This could give the startup a critical edge or make it an attractive acquisition target. Zelikman insists they’re not interested in being acquired: “We believe this is going to be a generational company, and we think that this has the potential to fundamentally change the future of how we interact with these models.”
The Resource Challenge: Computing Power as Currency
Success won’t come cheap. OpenAI’s recent financial disclosures reveal the staggering costs of AI development. The company’s annual revenue more than tripled to over $20 billion in 2025, driven by a massive expansion in computing capacity from 0.2 GW in 2023 to 1.9 GW in 2025.
OpenAI CFO Sarah Friar emphasizes: “Computing power is the scarcest resource in AI. Access to computing power determines who can scale.” For Humans&, this means competing with established players for the expensive compute resources needed to train and scale their coordination-focused model.
A New Paradigm for Business
The implications for businesses are profound. LinkedIn founder Reid Hoffman argues that companies are implementing AI wrong by treating it like isolated pilots. “The real leverage is in the coordination layer of work – how teams share knowledge and run meetings,” he writes. “AI lives at the workflow level, and the people closest to the work know where the friction actually is.”
This is exactly the space where Humans& wants to operate. Their model would act as “connective tissue” across organizations, understanding individual skills, motivations, and needs while balancing them for collective benefit. Whether they succeed could determine whether AI becomes truly integrated into how we work together or remains a collection of isolated tools.
The race is on to build AI that doesn’t just answer questions but helps us ask the right ones together. As businesses navigate this transition, the winners will likely be those who can harness AI not just for automation, but for enhancing human collaboration itself.