Hark previews browser-use agent for task completion

Hark, a startup that raised $700 million in Series A funding in May, today launched Hark Handoff, an agent that uses a browser efficiently to complete tasks. The company claims Handoff can easily navigate websites without official APIs, including Target, Walmart, OpenTable, and LinkedIn, by analyzing website structure and visual data to decide whether to click buttons or type information. Users can issue commands to order food or coffee, book travel, file returns, shop, reserve tables, or research topics. In a video demo, CEO Brett Adcock showed the assistant building a bouquet based on user specifications, including fuzzy terms like 'some of the florist's choice.' For this release, Hark uses a post-trained model and plans to pre-train later this year to refine its data pipeline and training infrastructure. Unlike LLMs that predict the next token, Hark's model predicts the next action, such as a click or keyboard input. Competitors include Google, OpenAI, Anthropic, and startups like Browser Use, Polar, Strawberry, and Aside. Hark claims Handoff is faster and costs less than models like GPT 5.5 and Opus 4.8. The startup has opened a waitlist and plans to release the agent by the end of summer.