Brett Adcock's Hark is not a $700 million venture round story, at least not on the public record. The verified story is still interesting: a self-funded AI lab trying to build the personal AI interface that smartphones have not yet become.
Hark has attracted attention because it sits at the intersection of three things investors still care about deeply: AI models, consumer hardware, and founders with a record of turning capital into ambitious companies. But the published version overstated the financing and mangled the copy so badly that the core story was almost unreadable.
The verified facts are more measured. Hark is a San Jose-based AI lab founded by Brett Adcock, the entrepreneur behind Figure AI and Archer Aviation. The Information reported that Adcock launched the company with $100 million of his own capital, based on a memo sent to Figure employees and investors. TechCrunch later reported that Hark is building multimodal models, hardware, software, and interfaces together, with the aim of creating a personal intelligence product rather than another chatbot wrapped inside an app.
That distinction matters. The consumer AI race has become crowded with companies trying to make artificial intelligence feel less like a text box and more like an active layer in daily life. Meta has pushed smart glasses. OpenAI is working with Jony Ive on AI-native hardware. Humane and Rabbit showed how difficult this category can be when the device arrives before the use case feels essential. Hark is entering the same territory, but with a different pitch: build the model and the interface as one system from the start.
The Vertical Integration Bet
Adcock's argument is that the current generation of AI tools is limited not only by model performance, but by the devices built around them. Phones, laptops, and apps were designed before generative AI became a daily utility. Hark wants to design around intelligence first, then decide what kind of hardware and interaction model should carry it.
That is why the company's hiring has drawn attention. Hark has brought in Abidur Chowdhury, a former Apple industrial designer who worked on recent iPhone design efforts, as director of design. TechCrunch reported that the company has about 45 engineers and designers, with talent from Apple, Meta, Tesla, and other major technology companies. For a young lab with no public product, that kind of team is the signal investors and competitors notice first.
The company's product remains mostly under wraps. Hark has said it is working on systems that can listen, see, understand context, and remember a user's life over time. It expects to release its first AI models this summer, with hardware to follow later. That schedule gives the market a near-term checkpoint, even if the bigger consumer hardware question will take much longer to answer.
Why Figure AI Matters
Hark is separate from Figure AI, but the connection is hard to ignore. Figure is building humanoid robots, and Hark's models are already being trained on Figure's robots, according to TechCrunch. That gives Hark an unusual source of embodied data, which could help its systems understand physical context better than models trained only on text, images, and screen-based interactions.
This is where the story becomes more than another AI device rumor. If Hark can use robotics data to build models that understand the world in a more practical way, it may have an advantage over consumer AI startups that depend entirely on cloud models and off-the-shelf hardware. That does not guarantee a useful product. It does make the experiment worth watching.
The funding picture needs caution. Public reports support Adcock's $100 million personal commitment, and private-market listings have shown Hark at multibillion-dollar valuation levels. But the published claim that Hark raised more than $700 million in a Series A led by Parkway Venture Capital could not be verified through meaningful current sources. That figure should not remain in the article unless supported by a primary announcement, regulatory filing, or credible reported source.
For founders, the takeaway is not that every AI startup should build hardware. Most should not. Hardware raises the cost of iteration, adds supply chain risk, and gives customers another device to reject. The real lesson is about product architecture. If AI is going to move beyond chat windows, the winning companies will need to think carefully about where intelligence lives, how users access it, and what the interface should do before a prompt is typed.
Hark is still an early bet with more ambition than proof. The next test comes when its first models reach users this summer. If they feel meaningfully different from today's assistants, Hark will have earned more attention. If not, it will join a growing list of companies that learned the hard way that replacing the smartphone is easier to pitch than to ship.
Also read: AMD challenges Nvidia with 4,000 local AI supercomputer • China's AI micro-drama boom is a business model worth watching • Agentic AI could outrun financial regulators
