AI Agents Northeastern and Stanford Give Agents a Git History: Shepherd Lifts Pair-Coding Pass Rates From 28.8 to 54.7 Percent Coding Assistants China Telecom Trains a 285B-Parameter Agentic Model on 8,192 Ascend 910C Chips With No Nvidia Silicon Anywhere in the Stack News Google Cuts 2027 Broadcom-Designed TPU 8i Orders by About a Million Units and Pushes Volume Toward MediaTek Business China's Data Regulator Will Set National Rules for Metering and Billing AI Tokens After Daily Use Tripled to 500 Trillion Business Shanghai's Yuanshi Technology Raises Nearly RMB 300 Million for AI Virtual Cells, Two Weeks After the Last Chinese Record Fell Security GAO Finds No Widely Accepted Test for Generative AI Cyber Skills and Hands Congress 16 Options, Including Human Sign-Off on Consequential Actions News Microsoft Turns Three USB4-Cabled PCs Into One AI Machine: Qwen3-235B Runs at Up to 22 Tokens a Second Across 384GB of Ordinary Memory Business Dell, HPE, Lenovo and Supermicro Are Demanding Up to 45% More for Signed AI Server Deals, and Trading Firms Are Paying AI Agents Northeastern and Stanford Give Agents a Git History: Shepherd Lifts Pair-Coding Pass Rates From 28.8 to 54.7 Percent Coding Assistants China Telecom Trains a 285B-Parameter Agentic Model on 8,192 Ascend 910C Chips With No Nvidia Silicon Anywhere in the Stack News Google Cuts 2027 Broadcom-Designed TPU 8i Orders by About a Million Units and Pushes Volume Toward MediaTek Business China's Data Regulator Will Set National Rules for Metering and Billing AI Tokens After Daily Use Tripled to 500 Trillion Business Shanghai's Yuanshi Technology Raises Nearly RMB 300 Million for AI Virtual Cells, Two Weeks After the Last Chinese Record Fell Security GAO Finds No Widely Accepted Test for Generative AI Cyber Skills and Hands Congress 16 Options, Including Human Sign-Off on Consequential Actions News Microsoft Turns Three USB4-Cabled PCs Into One AI Machine: Qwen3-235B Runs at Up to 22 Tokens a Second Across 384GB of Ordinary Memory Business Dell, HPE, Lenovo and Supermicro Are Demanding Up to 45% More for Signed AI Server Deals, and Trading Firms Are Paying

Hark Pro Goes Wide: Brett Adcock's $6 Billion AI Startup Opens a Free Assistant That Runs Your Computer

Hark Pro Goes Wide: Brett Adcock's $6 Billion AI Startup Opens a Free Assistant That Runs Your Computer

Hark, the startup Brett Adcock founded after Figure AI and Archer Aviation, has opened wide release of Hark Pro, a computer-use assistant that operates a user's browser, email and calendar. The product is free with $20 and $100 usage tiers, and the company says purpose-built AI hardware arrives in 2027.

Hark, the startup Brett Adcock founded less than a year ago, has opened wide release of Hark Pro, an AI assistant built not to chat but to operate a computer. Access is free, with paid subscription tiers for heavy users, and the company frames the product as an interface for AI rather than a step toward artificial general intelligence.

The model underneath is trained specifically for computer use, which Hark says makes it faster and cheaper at exactly those tasks. The experience is a full-screen workspace with a central chat dialog, a feed of suggested actions, and panels that act as small dashboards for topics the user cares about. Even the background imagery adapts to the local weather and time of day.

In a demo with TechCrunch, design lead Abidur Chowdhury, who previously worked at Apple, showed the assistant proactively surfacing expenses that needed approval, emails awaiting a reply and an upcoming flight it offered to book. He also described a case where it recognized that a colleague needed to renew a car registration with the California DMV and completed the process through the agency's website without a handoff.

One deliberate design choice stands out: Hark shows the user, in a small window, how the agent navigates the web, an action most assistants perform invisibly. That transparency ties into the company's privacy pitch. "We're not here to sell you ads and steal your data," Chowdhury said, adding that the system is built to understand tasks rather than to profile users or the people around them.

Reporting on the launch describes a three-tier structure: a free plan, a $20 per month tier with double usage, and a $100 per month tier with ten times the allowance. Hark Pro runs on Handoff, the cloud infrastructure the company introduced in August, which coverage says can drive up to 36 browser sessions in parallel and manage tasks across millions of websites.

The company behind it has moved quickly. Adcock seeded Hark with $100 million of his own capital before it emerged from stealth in March 2026, then raised more than $700 million in a Series A on May 21 at a $6 billion post-money valuation, led by Parkway Venture Capital with Nvidia, AMD Ventures, Intel Capital, Qualcomm Ventures, Salesforce Ventures and ARK Invest participating. The team is around 70 people and currently lists 55 open roles.

Hark Pro lands roughly four weeks after Meta launched Muse, a proactive assistant with a similar tiered structure, and enters a field that already includes Meta's Muse, Instinct and OpenAI's Dots. The differentiator Adcock's team claims is completion: "It might not be faster than Muse, but it will give you a better result, and it will work end-to-end instead of asking you to take over at a point," Chowdhury said.

Software is only half the plan. Hark has promised AI-native hardware in 2027, with AT&T lined up as a connectivity partner, and has ruled out wearable cameras, which Chowdhury called "creepy." Whether the category consolidates around one product template before any of that hardware ships, or splinters by ecosystem, is now the open question for the assistant market.

Comments (0)

Log in to join the discussion

Log In

No comments yet