Imagine a world where AI doesn’t just answer questions or crunch numbers, but seamlessly orchestrates human collaboration, navigating the messy realities of competing priorities and long-term goals. That’s the bold vision driving Humans&, a new startup founded by ex-Anthropic, Meta, OpenAI, xAI, and Google DeepMind alumni. While AI chatbots excel at individual tasks, they’re still far from mastering the art of teamwork—a gap Humans& aims to close with a groundbreaking approach to foundation models.
But here’s where it gets controversial: Humans& isn’t just tweaking existing AI architectures; they’re building something entirely new—a model designed for social intelligence. This isn’t about retrieving information or generating code; it’s about understanding human dynamics, motivations, and needs to facilitate collaboration on a scale we’ve never seen before. This week, the company secured a staggering $480 million seed round (https://techcrunch.com/2026/01/20/humans-a-human-centric-ai-startup-founded-by-anthropic-xai-google-alums-raised-480m-seed-round/) to bring this vision to life, positioning itself as the “central nervous system” of the human-plus-AI economy.
And this is the part most people miss: Humans& isn’t just another productivity tool. Co-founder Andi Peng, a former Anthropic employee, describes it as the dawn of AI’s second wave, where consumers move beyond treating AI as a smart assistant and start leveraging it as a collaborative partner. “We’re trying to figure out what to do with all these things,” Peng told TechCrunch, highlighting the shift from vertical-specific models to systems that empower humans to work together more effectively.
The startup’s pitch is timely. As companies transition from chat-based AI to agent-driven workflows, the coordination challenge remains largely unaddressed. Models are competent, but workflows aren’t—and people feel both threatened and overwhelmed by AI. Humans& aims to bridge this gap, though it’s still early days. The three-month-old company has no product yet, but its founders hint at a platform that could replace multiplayer contexts like Slack (https://techcrunch.com/2026/01/13/slackbot-is-an-ai-agent-now/) or Google Docs, catering to both enterprise and consumer needs.
Co-founder and CEO Eric Zelikman, a former xAI researcher, emphasizes the focus on communication and collaboration. “We’re building a product that helps people work together and communicate more effectively—both with each other and with AI tools,” he explained. For instance, imagine making a group decision, like choosing a logo, without the tedious back-and-forth. Humans&’s model would act as a mediator, asking questions in a way that feels natural, almost like a friend or colleague trying to understand your perspective.
But here’s the kicker: Unlike chatbots that ask questions without context, Humans&’s model will be trained to understand the value of those questions. It’s not just about immediate user satisfaction or correct answers; it’s about fostering meaningful collaboration. This requires rethinking how AI models are trained, incorporating long-horizon and multi-agent reinforcement learning (RL) to enable planning, revision, and coordination over time.
This approach is gaining traction in academic circles (https://arxiv.org/pdf/2502.01600), but Humans& is taking it a step further by designing the product and model in tandem. “As the model improves, we’re able to co-evolve the interface and behaviors into a product that makes sense,” Peng explained. The goal? To own the collaboration layer, not just integrate into existing tools.
But is this vision too ambitious? Humans& isn’t just competing with collaboration tools like Notion or Slack; it’s going head-to-head with AI giants like Anthropic, Google, and OpenAI. These companies are already enhancing collaboration on their platforms, and some argue that AGI will eventually replace human work. LinkedIn founder Reid Hoffman recently pointed out that the real leverage lies in the coordination layer—how teams share knowledge and run meetings (https://x.com/reidhoffman/status/2014028797507449206?s=61). That’s exactly where Humans& wants to play, acting as the “connective tissue” across organizations, balancing individual skills and motivations for the greater good.
Yet, the risks are immense. Training and scaling a new model requires endless resources, putting Humans& in direct competition with established players for compute power and talent. And with tech giants like Meta and OpenAI scouting for acquisitions, there’s always the risk of becoming a target. Humans& insists it’s here to stay, with Zelikman declaring, “We believe this is going to be a generational company.”
So, here’s the question for you: Can Humans& truly redefine how we collaborate with AI, or is it biting off more than it can chew? Will its focus on social intelligence give it an edge, or will it fall prey to the very giants it aims to challenge? Let us know your thoughts in the comments—this is one conversation you won’t want to miss.
