Published: October 6, 2026 · Last updated: October 6, 2026
TL;DR: Hark, Brett Adcock’s AI startup that raised $700 million at a $6 billion valuation earlier this year, released its first product today. Hark Pro is a free, full-screen AI assistant built specifically for computer-use tasks rather than general chat, and the company is selling it on privacy at the exact moment its biggest rival is facing backlash over its data terms.
Hark spent most of this year as one of those startups everyone in AI circles talked about and nobody could actually use. Today that changed. The company released Hark Pro, a free AI assistant with a paid tier for heavier users, and it’s the first real product to come out of the $700 million Series A that valued Hark at $6 billion back in the spring.
What makes Hark Pro different from the dozen other things people are calling an “AI assistant” this year is what it was actually trained to do. It isn’t a wrapper around a frontier chatbot and it isn’t trying to win a benchmark leaderboard. It’s a full-screen, operating-system-style interface trained specifically for computer-use tasks, meaning it watches what’s on your screen and acts on it directly instead of waiting for you to describe your problem in a chat box. That’s a narrower bet than building a general-purpose model, and it’s also a much harder one to fake with a good demo.
Brett Adcock, who’s built a reputation for jumping into hardware-adjacent AI bets and raising absurd sums to do it, is positioning Hark Pro as a stepping stone rather than the finished product. Abidhur Chowdhury, Hark’s VP of design and a former Apple designer who worked on the iPhone Air, has said the company is building toward a dedicated AI hardware device targeted for 2027. The software launching today is as much about getting real usage data into that hardware roadmap as it is about winning assistant users right now.
The timing of Hark’s privacy pitch is not an accident. Hark Pro is entering a field that already includes Meta’s Muse, a startup called Dots, and Instinct, the personal AI agent company that just raised $1 billion at a $10 billion valuation with a grand total of 14 employees. Instinct’s news made headlines for the valuation, but it’s also drawn a separate round of criticism over the perpetual-license terms buried in how it handles user data once its agent is let loose in group chats. Hark is walking directly into that opening, built around the explicit promise that it isn’t going to treat your screen activity as training material it owns forever.
Whether that promise means anything in practice is the real question, and it’s one Hark hasn’t answered yet with specifics on data retention or what “free” actually costs users in the fine print. An assistant that’s useful because it watches everything on your screen and a company that’s trustworthy because it promises not to exploit that access are two claims that are structurally in tension, and no company in this category, Hark included, has shipped anything yet that proves both can be true at once. Positioning yourself as the privacy-first option is cheap. Earning it over years of actually handling people’s screens is not, and that’s the test Hark just signed up for in public.
Related: Instinct just raised $1 billion at a $10 billion valuation with 14 employees and a 19-year-old raised $11 million to sell a computer that only runs AI.
Bottom Line: Hark picked the right moment to make privacy its pitch, with Instinct catching heat over its data terms in the same week. But a promise is not a policy, and the company that actually proves it can watch your screen without hoarding what it sees is going to win this category, not the one that says it first.
Follow Teck Hustlers: Facebook · X · Instagram · LinkedIn · TikTok · Pinterest
