Hark Releases Handoff: A Faster AI Agent for Computer Use in Hark Pro

·
·
2 min read
·
AI-assisted
Author Profile
by Albert Schaper
Share
Hark Releases Handoff: A Faster AI Agent for Computer Use in Hark Pro

Hark has released Handoff, the second generation of its computer-use agent for Hark Pro, which operates web tasks roughly 2x faster than its predecessor by directly interacting with live pages through clicking, typing, and scrolling.

Enhanced Performance and Speed

According to Hark's internal benchmarks, Handoff demonstrates notable improvements in task completion and speed. On a 478-task internal benchmark, the agent completes approximately 82% of tasks. This represents a roughly 2x increase in speed for web tasks compared to its predecessor, version 1.

The speed enhancements are attributed to the implementation of reinforcement learning, which was designed to penalize wasted turns and reduce latency during operations. Specific improvements include flight booking tasks running up to 3.8x faster and long research tasks completing more than 5x faster. On average, Handoff completes web tasks about 2x faster than version 1.

Previous Version's Impact and Current Availability

The first iteration of Handoff, launched in Hark Pro in October, has already processed over 5.6 million browsing sessions and facilitated more than $20 million in purchases. The current Handoff agent runs on Hark's cloud infrastructure, providing a dedicated virtual computer for each request. It is accessible across multiple platforms, including web, iOS, and Android.

Comparison with Other AI Agents

While Handoff shows significant internal improvements, Hark acknowledges that its browsing completion still trails some competitors. On the same internal harness, Handoff lags behind Anthropic's Claude Opus 5.5 (max) by up to 5 points in browsing completion. This comparison highlights the ongoing competitive landscape in the development of AI agents for computer use.

Key Technical Details

Handoff's development involved co-evolving its reinforcement learning model with a new agent harness. This approach specifically targets the reduction of inefficient actions and delays, contributing to the observed performance gains. The agent's ability to directly interact with web elements, rather than relying on underlying APIs, is a core aspect of its operational design.

Conclusion

Hark's release of Handoff marks a significant update to its computer-use agent, focusing on increased speed and efficiency for web-based tasks within the Hark Pro app. The reported 2x speed increase and improved task completion rates, driven by reinforcement learning, position Handoff as a more capable tool for automated browsing. While it demonstrates strong internal performance, the company continues to refine its capabilities in comparison to other advanced AI models in the market.

Sources

About the Author

Albert Schaper avatar

Written by

Albert Schaper

Albert Schaper is a co-founder of Best-AI.org. He focuses on product strategy, AI adoption, practical tool selection, and educational content that helps users compare AI products with clearer context.

More from Albert

Was this article helpful?

Found outdated info or have suggestions? Send us a note.

Discover AI Tools

Find your perfect AI solution from our curated directory of top-rated tools

Less noise. More results.

One monthly email with the product launches tools that matter - and why.

No spam. Unsubscribe anytime. We never sell your data. See our Privacy Policy.

What's Next?

Continue your AI journey with our tools and resources. Whether you're looking to compare AI tools, learn about artificial intelligence fundamentals, or stay updated with the latest AI news and trends, see what fits your needs. Explore our curated content to find the right AI tools for your workflow.