Humans& thinks coordination is the next frontier for AI, and they’re building a model to prove it

AI chatbots have made significant strides in areas such as answering queries, summarizing documents, and solving equations. However, they are still primarily designed to assist individual users, lacking the capability to manage the complexities of true collaboration, which involves coordinating teams with differing priorities, tracking ongoing decisions, and ensuring sustained alignment among members. Enter Humans&, a startup founded by former employees from top tech companies including Anthropic, Meta, OpenAI, xAI, and Google DeepMind. The company is focused on bridging this gap, viewing it as the next major challenge for foundational AI models. Recently, Humans& secured a remarkable $480 million in seed funding aimed at creating a “central nervous system” for the human-AI collaboration landscape. While early coverage has emphasized their goal of empowering individuals with AI, the startup’s more innovative ambition is to design a new model architecture that prioritizes social intelligence over mere data retrieval or code generation. Andi Peng, a co-founder and former Anthropic team member, shared with TechCrunch that we are transitioning from a phase of scaling specialized question-answering models to a new era where everyday users are trying to navigate the multitude of AI tools at their disposal. Humans& envisions a future where AI enhances rather than replaces human jobs, a message that is particularly relevant as organizations move from simple chatbots to more sophisticated agents. Despite the competence of current AI models, workflow issues remain unresolved, leaving many feeling overwhelmed by AI advancements. Although the company is still in its infancy and has yet to launch a product, they are exploring possibilities that could replace existing communication platforms like Slack or collaboration tools such as Google Docs and Notion. The team hints at catering to both enterprise and consumer markets, with an emphasis on creating a product that facilitates effective communication and teamwork. Eric Zelikman, co-founder and CEO, emphasized that their model will be designed to engage users in a friendly manner, fostering a collaborative atmosphere. Unlike current chatbots that ask questions without real understanding, the new model aims to be more intuitive and relational. This approach is part of a broader strategy to evolve the interface in tandem with the model's development. Humans& doesn’t plan to simply create a new AI model that integrates with existing tools; instead, they aim to own the collaboration space. This ambition aligns with a growing trend in the AI sector, where startups like Granola, a note-taking app, have also raised significant funds to enhance their collaborative features. Thought leaders, including LinkedIn founder Reid Hoffman, stress the importance of the coordination layer in AI, arguing that success lies in how teams leverage knowledge and manage meetings. Humans& aims to be that connective tissue within organizations, understanding the skills and needs of individuals while balancing them for collective benefit. To achieve their vision, Humans& is rethinking AI model training, focusing on collaborative interactions between humans and AIs. Co-founder Yuchen He highlighted the use of long-horizon and multi-agent reinforcement learning approaches to help the model plan and adapt over time, moving beyond simplistic, one-off responses. However, the path ahead is fraught with challenges. Training and scaling a new model requires substantial funding, putting Humans& in direct competition with established players for resources. Moreover, they are not only up against collaboration platforms like Notion and Slack but also major AI companies that are enhancing their own collaborative capabilities. Despite the competitive landscape, Humans& has expressed confidence in its long-term vision and has already declined offers for acquisition. Zelikman confidently stated, “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.”