Hark, the AI startup that raised $700 million in Series A funding in May, today launched Handoff, a browser-use agent designed to complete tasks across websites that lack official APIs. The company says Handoff can work through sites like Target, Walmart, OpenTable, and LinkedIn by interpreting page structure and visual cues to decide where to click or type.
The agent is part of a crowded field of browser-based automation tools, but Hark claims its approach differs by predicting the next action rather than the next token. In a demo, CEO Brett Adcock showed Handoff assembling a flower bouquet based on user specifications, including fuzzy instructions like “some of the florist’s choice.” However, the video only showed part of the process, leaving questions about real-world reliability.
Also read: Anthropic assembles in-house team to design custom AI chips
How Handoff works and what it promises
Handoff operates by issuing a command — such as ordering coffee, booking travel tickets, filing tax returns, or researching a topic — and then executing the steps in a browser. Hark says it uses a post-trained model for this release, with plans to pre-train later this year to refine its data pipeline and training infrastructure.
The company emphasizes speed and cost advantages over competitors. Hark claims Handoff is faster and significantly cheaper than models like GPT-5.5 and Opus 4.8, though it did not provide specific benchmarks in the announcement. The startup has opened a waitlist and expects to release the platform by the end of summer 2026.
Competitive sector and market context
Hark is entering a market already occupied by major players. Google, OpenAI, and Anthropic have all developed computer-use agents, and VC-backed startups like Browser Use, Polar, Strawberry, and Aside are working on similar browser automation. The key differentiator for Hark is its claimed efficiency and pricing, which could appeal to consumers and businesses looking for cost-effective task automation.
The $700 million Series A, announced in May, signals strong investor confidence in the space, but it also raises expectations for Handoff to deliver on its promises. The company’s focus on post-training first, then pre-training, suggests a deliberate strategy to iterate quickly on user feedback before scaling up.
What to watch for
As Handoff moves toward its public release, observers will be watching for independent evaluations of its performance, particularly on complex tasks that require handling dynamic web pages and unexpected pop-ups. Hark’s claims about cost and speed will also be tested against real-world usage.
The broader trend of AI agents handling routine online tasks is gaining momentum, but reliability remains a hurdle. If Handoff can deliver on its promises, it could carve out a niche in the consumer and small-business market. For now, the waitlist will give early adopters a chance to see if the agent lives up to the hype.
Disclaimer: This article discusses a product launch and its claimed capabilities, including performance and pricing comparisons. These claims are forward-looking and unverified by independent testing. The information provided is for editorial purposes only and does not constitute financial or investment advice. The AI agent market is volatile and evolving, and actual performance may vary.
