Skyvern 2.0

Build AI browser agents in plain english

Screenshots

Hunter's comment

Skyvern is an open source Browser Agent Builder. You can build complex browser agents in plain english. Skyvern scores a state-of-the-art 85.8% on the WebVoyager benchmark, enabling companies to build automations for job apps, contact form filling, and more.
Check out Skyvern cloud (https://app.skyvern.com) or open source (https://github.com/Skyvern-AI/Sk...).

Skyvern 2.0 brings state-of-the-art performance and scores 85.8% on the WebVoyager benchmark. We achieved this by building an agent with a self-reflection loop allowing it to break complex tasks down into smaller pieces, and validate that each piece is actually solved correctly. The full technical report can be found here: https://blog.skyvern.com/skyvern...

❓How does Skyvern work?

Skyvern operates by being given a URL + goal-based prompt. Skyvern then takes actions on a website until it accomplishes the goal given to it. We use multi-modal LLMs under the hood to parse the viewport and interact with it the way a human would.