Use a browser copilot to complete PDF forms

A browser-based agent can pull details from connected records or typed input and place them into PDF fields for review before signing.

Done with Custom agent· Intermediate· May 2026

What they did
A custom copilot ran inside a PDF editor in the browser and used tool calls to work through form fields. It could take information from a connected source like a CRM or EHR through MCP or RAG, or from text the user entered. As it filled each field, the edits showed up right away, and the user could fix any entry with a single click. It could also help with forms in another language and point out contract language before a signature.
How it went
The source did not give usage numbers. A commenter later spotted a bug where corrupted field labels were passed to the model, which led to values going into the wrong boxes; that problem has since been fixed.
Tips
Keep the labels sent to the model clean and structured, or field mappings can break. Make the agent stop for review before any submit, send, or sign action.

Try it yourself with Claude in Chrome

Open the PDF at [URL or file] and help me complete it using these details: [paste name, address, dates, and other info]. If the form uses [language], translate each field for me and explain any clause I should understand before I sign. Fill the fields one by one, show me each completed field for approval, and pause before any submit, send, delete, payment, or signature step.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on news.ycombinator.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.