In a striking sign of the continued frenzy surrounding artificial intelligence investments, the start-up Humans& has announced it raised $480 million (£357 million) in a funding round that values the three-month-old company at $4.48 billion. The round, which included heavyweights such as Nvidia, Amazon founder Jeff Bezos, and venture capital firms SV Angel and Google Ventures, underscores investors' relentless appetite for innovative AI companies, even those that have yet to ship a product.
Humans&, which has only about 20 employees, is still in stealth mode with no product publicly released. Yet its valuation — more than $200 million per employee — reflects the market's bet on the founding team's vision and track record. The company operates in the domain of human-centric AI, a subfield that focuses on building artificial intelligence systems designed to collaborate with humans rather than replace them. This approach stands in contrast to many current AI products that aim for full automation—such as chatbots that handle customer service or algorithms that manage supply chains without human intervention.
What Is Human-Centric AI?
Human-centric AI, sometimes called collaborative intelligence, emphasises augmenting human capabilities. The idea is not to build systems that work independently, but ones that can partner with teams of humans, ask clarifying questions, store information for later retrieval, and evolve into ongoing resources that adapt to a group's workflow. For example, an AI model could participate in a project meeting, note action items, and later remind team members of deadlines or suggest relevant data from past projects.
This philosophy is gaining traction in industries where tight human oversight remains critical, such as medicine, law, and creative fields. In healthcare, a human-centric AI might assist a doctor by analysing scans but then present its findings in a conversational style, inviting the physician to ask for more details or override its suggestions. In software development, such an AI could work alongside engineers to debug code, ask for clarification on ambiguous requirements, and learn from each interaction.
The founders of Humans& bring deep expertise from the frontier of AI research. Georges Harik, one of the company's co-founders, was Google's seventh employee and a key figure in the early development of Gmail, Google Docs, and the acquisition of Android. Harik also played a major role in shaping the company's cloud and advertising infrastructure. His experience at Google spanned a period of explosive growth that saw the company become a global technology powerhouse.
Eric Zelikman, the CEO and co-founder, previously worked at xAI, Elon Musk's artificial intelligence venture, where he contributed training data for the Grok-2 chatbot. Zelikman also has a background in reinforcement learning, specifically focusing on methods that teach AI systems to reason through problems using rewards — a technique that could be crucial for building AIs that engage in long-term collaborative reasoning with humans. The team also includes researchers from Anthropic and OpenAI, two of the most prominent AI labs known for their work on large language models and safe AI.
The Investment Landscape
The funding round was led by some of the same names that have dominated AI investing in 2024 and 2025. Nvidia, the GPU maker whose chips power most AI training and inference, has become one of the largest corporate investors in AI startups, placing strategic bets on companies that could drive demand for its hardware. Jeff Bezos, through his family office, has been an active angel investor in a range of AI ventures, including Perplexity AI and Skydio. Google Ventures, the venture capital arm of Alphabet, also participated, reflecting a continued corporate interest in nurturing early-stage AI companies that align with the parent company's cloud and AI services.
SV Angel, a well-known early-stage fund that has backed Coinbase, Airbnb, and now Humans&, brings a network of relationships in Silicon Valley. The involvement of these heavyweight investors is a clear signal that they believe in the premise of collaborative AI — and that they see a massive market opportunity in making AI systems work with people, not just automate tasks away.
Humans& is operating in a crowded field. Major tech companies like Microsoft, Google, and Meta are all pouring resources into AI assistants that can interact with humans. But Humans& differentiates itself by focusing specifically on team collaboration rather than individual assistant tasks. For instance, while many chatbots are designed for one-on-one interactions, the AI envisioned by Humans& would be able to hold a conversation with multiple participants, keep track of different roles and responsibilities, and store information hierarchically for reuse across projects.
The concept is reminiscent of earlier research in multi-agent systems, but applied to real-world business contexts. Many organisations still struggle to adopt AI because employees feel overwhelmed by tools that are either too passive or too intrusive. A human-centric AI that can ask for consent, learn from feedback, and gradually embed itself into a team's culture could break down adoption barriers.
Historical Context and Market Rationale
The huge valuation at this nascent stage is reminiscent of the dot-com era, but the startup may justify it through the sheer size of the addressable market. According to estimates from McKinsey, AI-powered collaboration tools could add up to $2.6 trillion in annual economic value by 2030, especially in knowledge work sectors such as finance, legal, and consulting. Investors are betting that Humans& will capture a meaningful share of that market.
The company's choice of name — Humans& — is deliberately symbolic: the ampersand highlights the partnership between humans and AI. This branding reflects a broader shift in the industry's rhetoric. Just a few years ago, many AI companies marketed their products as replacements for human workers. Today, largely due to public backlash and regulatory pressure, the tone has changed. Terms like "augmented intelligence," "collaborative AI," and "human-in-the-loop" are now common. Humans& is riding this wave.
The three-month-old firm has not disclosed specific product plans, but job postings indicate they are hiring for roles in reinforcement learning, natural language processing, and full-stack engineering. A typical day at the company might involve tweaking a large language model to better follow conversation threads, then testing it with a simulated team that includes both human-like agents and real employees. The goal is to create an AI that can schedule meetings, draft emails, compile reports, and even mediate disagreements — all while respecting human authority.
One challenge for the startup is the intense competition for talent. The top AI researchers are currently commanding salaries in the millions of dollars, and companies like OpenAI, Anthropic, and Google DeepMind are constantly poaching from each other. Humans& will need to leverage its founding team's reputation to attract world-class engineers. The involvement of Nvidia and Google Ventures could also provide access to cloud credits and hardware, which are essential for training large models.
Another challenge is regulation. The European Union's AI Act, which came into force in 2024, imposes strict requirements on high-risk AI systems, including those that interact with humans. While human-centric AI that collaborates rather than autonomously makes decisions may be subject to fewer restrictions, the company will still need to ensure transparency, privacy, and the ability to opt out. The founding team's experience at top AI labs, which have been at the forefront of safety research, should help them navigate these legal waters.
Expanding the Team and Vision
Georges Harik, who also served as one of the leaders of the funding round, has a background not only in Google but also in several startups. After leaving Google, he co-founded a number of ventures, including the data analytics firm Kaggle (later acquired by Google) and the robotics company Kindred. His deep understanding of both the technical and business sides of AI makes him a valuable asset for Humans& as it scales.
Eric Zelikman's work on reinforcement learning at xAI is particularly relevant. Reinforcement learning (RL) is a branch of machine learning where an agent learns by taking actions in an environment to maximise a reward signal. In the context of human-AI collaboration, RL can be used to teach an AI to adapt its behavior based on feedback from human teammates. For instance, if a team's AI suggests a meeting time that conflicts with most people's schedules, it should learn to avoid that. Over time, it could optimise for team productivity without explicit programming.
Zelikman also contributed to and is known for the "Quiet-STaR" technique, a method that trains language models to generate internal reasoning before responding. This could be crucial for making AI more thoughtful and less prone to hallucination when collaborating.
The company is also likely benefiting from the exodus of talent from larger firms. In 2024, several high-profile researchers left OpenAI and Anthropic to start their own ventures, citing a desire to explore more creative or less restrictive research directions. Humans& is among a handful of new startups that have emerged from this brain drain, alongside others like Sentient and Cohere. The result is a rich ecosystem of competing approaches to AI safety and alignment.
Market Reactions and Future Outlook
Following the funding announcement, analysts have been quick to comment. Some note that a $4.48 billion valuation for a company with no revenue and only 20 employees is a stark reminder of the speculative nature of AI investments. Others argue that, given the potential of AI to reshape every industry, early bets on visionary teams can produce outsized returns. The fact that the company is focusing on a niche — team collaboration — rather than trying to outdo generalist chatbots like ChatGPT may actually help it avoid direct competition with giants.
There are also risks. The human-centric AI space is already attracting copycats. Startups such as Cohere for Business and People+AI are developing similar concepts. Moreover, the success of Humans& heavily depends on execution. Building an AI that truly understands human nuance — such as sarcasm, hesitation, or non-verbal cues through text — remains unsolved. The company has not yet published any research papers or demos, so the public has little to judge beyond the founders' reputations.
Nevertheless, the funding round signals strong confidence that the team can deliver. The investors in this round are not typical early-stage speculators; they are seasoned players with deep industry knowledge. Nvidia, for example, rarely invests in small startups unless they are strategically aligned with its hardware ecosystem. By backing Humans&, Nvidia is betting that collaborative AI will require massive computational resources for both training and inference — resources that only Nvidia's GPUs can efficiently provide.
Jeff Bezos, meanwhile, has been increasingly interested in AI that augments human decision-making. His other investments in the sector, such as Perplexity AI (an AI-powered search engine), indicate a belief that the future of AI lies not in replacing human workers but in providing them with superpowers. Humans& fits squarely into that thesis.
As the company grows, it will need to hire rapidly, develop its core product, and build a customer base. The typical timeline from seed funding to product launch for AI startups is 12 to 18 months. Given the three-month lifespan, we can expect a beta version within the next year. The company's website currently shows only a landing page with a mailing list sign-up, suggesting a tight-lipped strategy until they are ready to unveil their work.
For now, the AI industry watches with bated breath. If Humans& succeeds, it could redefine how organisations integrate artificial intelligence into their daily operations. Instead of fearing that AI will replace them, workers might welcome it as a useful teammate. That shift in perception alone could unlock huge efficiencies and drive broader adoption across sectors that have so far been hesitant, such as government, education, and traditional manufacturing.
The trajectory of Humans& will also be watched by regulators, who are grappling with how to govern AI systems that interact closely with humans. If the company can demonstrate that its approach reduces bias, increases transparency, and respects user agency, it could set a positive precedent for the entire industry.
In the broader context, the $480 million raise is another data point in a year of record AI investment. According to PitchBook, global AI startup funding reached $98 billion in the first nine months of 2024, on pace to exceed $130 billion for the full year. The lion's share has gone to infrastructure companies and large model providers, but a growing portion is trickling down to application-layer companies like Humans&. This indicates a maturation of the ecosystem, where investors are looking for real-world value rather than just raw model capabilities.
Humans& may still be a toddler in the corporate world, but with its deep pockets, formidable founders, and compelling vision, it has the ingredients to become a major player in the next wave of artificial intelligence.
Source: Silicon UK News