Show HN: Filling PDF forms with AI using client-side tool calling
Not a pilot. SimplePDF Copilot lets you fill PDFs with AI—without ever sending your document to the cloud.
Why it matters
A developer shipped client-side tool calling for PDFs, solving a real privacy problem for healthcare teams. This is the emerging pattern: AI features that keep sensitive data in the browser, not in vendor clouds.
The key facts
17 to knowSimplePDF Copilot: AI assistant for PDF field filling, form completion, and document manipulation
All parsing, rendering, and field detection runs client-side—PDFs never leave the browser
Built on SimplePDF, used by 200k+ monthly users, 7-year-old project
Client-side tool calling via iframe postMessage architecture
Privacy model: LLM sees only necessary text/messages; PII stays local
Supports BYOK (bring your own key) and local LLM inference (tested with LM Studio)
Default LLM: DeepSeek V4 Flash (demo proxy)
Tech stack: Tanstack Start, Vercel AI SDK, Tailwind
Open source demo on GitHub; SimplePDF core remains proprietary
Solves healthcare compliance need for document privacy without third-party shipping
SimplePDF Copilot: AI assistant for PDF form filling, field detection, page management—runs client-side
200k+ monthly users on SimplePDF base product (7-year-old privacy-first PDF editor)
Client-side tool calling architecture: LLM intent + browser execution (speed + data containment)
Supports bring-your-own-key (BYOK) to any cloud LLM or fully local inference (LM Studio tested)
Default model: DeepSeek V4 Flash via demo proxy
Tech stack: Tanstack Start, Vercel AI SDK, Tailwind; open-source demo on GitHub
Use case: Healthcare/regulated industries where PII/PHI cannot leave the browser
Go to the source
Hacker Newscopilot.simplepdf.com
Publisher excerpt: Hey HN! I built SimplePDF Copilot: an AI assistant that can interact with the PDF editor. It fills fields, answers questions, focuses on a specific field, adds fields, deletes pages, and so on. It's built on top of SimplePDF that I started 7 years ago, pioneering privacy-respecting client-side pdf…