Anthropic Adds Computer-Use API to Claude 3.5 Sonnet
Anthropic released an API endpoint that lets Claude 3.5 Sonnet move the mouse, click, type, and scroll within a virtual desktop. The model executes multi-step workflows such as spreadsheet updates and web form submissions without requiring users to write code. Early testers report completing 30-minute data-entry tasks in under 5 minutes.
Screen-control capability converts language models from text generators into workflow agents. Users must now design explicit task sequences and guardrails rather than single prompts. This changes project scoping from API-call counts to end-to-end process ownership.
Ramp's engineering team integrated the computer-use API to automate receipt categorization across QuickBooks and Slack, reducing manual review time from 18 hours per week to under 3 hours while maintaining 97 percent accuracy on 1,200 weekly transactions.
Step 1: Sign up at anthropic.com and enable the computer-use beta in your account settings. Step 2: Install the Anthropic Python SDK and create a session that grants the model access to a sandboxed desktop environment. Step 3: Send a structured task prompt such as 'Open Excel, paste the CSV from this URL, and format column B as currency,' then review the recorded action log for errors.