The AI PDF editor your agent can operate
Paste one sentence. Watch the agent fill a W-9.
This is Copilot, our MIT-licensed demo built on the SDK. Open it, paste the prompt below into the chat, watch the fields fill, then correct one yourself. Use sample data only: the demo sends chat to a SimplePDF demo server unless you add your own AI key.
Fill this W-9 for Daniel Ostervane, individual/sole proprietor, 418 Vantmoor Dr, Austin, TX 78745Try the live demoSame PDF. Your agent does the typing.
Today
5 stepsYour users do the typing.
- Your app has the answers, the PDF doesn't
- Your users retype them by hand
- They guess at confusing fields
- One wrong field, the form comes back
- Or you build your own PDF backend
With SimplePDF
3 stepsYour agent does the typing. Your users press Submit.
- Embed the editor in your app
- Your agent fills the PDF
- Your users check, sign and press Submit
A few lines. No PDF backend to run.
Add the tools to your Vercel AI SDK or TanStack AI chat. They run in the browser, next to the PDF. Your chat route and your model stay yours.
Read the docsimport { EmbedPDF } from '@simplepdf/react-embed-pdf';
export function Document() {
return (
<EmbedPDF
mode="inline"
style={{ width: 900, height: 800 }}
companyIdentifier="yourcompany"
document={{ url: 'https://cdn.simplepdf.com/simple-pdf/assets/forms/fw9.pdf' }}
/>
);
}// <div id="editor" style="height: 100vh"></div>
import { createEmbed } from '@simplepdf/embed';
const embed = createEmbed({
// a CSS selector, or the HTMLElement itself
target: '#editor',
companyIdentifier: 'yourcompany',
document: { url: 'https://cdn.simplepdf.com/simple-pdf/assets/forms/fw9.pdf' },
});<!-- every link to a PDF on the page opens in your editor -->
<script
src="https://unpkg.com/@simplepdf/web-embed-pdf"
companyIdentifier="yourcompany"
defer
></script>
<a href="https://cdn.simplepdf.com/simple-pdf/assets/forms/fw9.pdf">
Fill the W-9
</a><!-- your branded editor, no JavaScript needed -->
<iframe
src="https://yourcompany.simplepdf.com/editor?open=https://cdn.simplepdf.com/simple-pdf/assets/forms/fw9.pdf"
width="100%"
height="800"
frameborder="0"
></iframe>import { useChat, fetchServerSentEvents } from '@tanstack/ai-react';
import { EmbedPDF, useEmbed } from '@simplepdf/react-embed-pdf';
import { useEmbedTools } from '@simplepdf/react-embed-pdf/tanstack-ai';
export function AgentEditor() {
const { embedRef } = useEmbed();
const tools = useEmbedTools(embedRef);
// the model's tool calls run against the open PDF, in the browser
useChat({ connection: fetchServerSentEvents('/api/chat'), tools });
return (
<EmbedPDF
ref={embedRef}
mode="inline"
companyIdentifier="yourcompany"
style={{ width: 900, height: 800 }}
/>
);
}Your AI provider, with no middleman
The editor runs in the browser, and your app calls the AI provider you choose.
How it works
- The PDF is opened, filled and signed in the browser
- AI requests carry the document context they need, straight to your provider
- With your own storage, submissions go to your S3 or Azure Blob Storage (Pro) or SharePoint (Premium)
What this means for your data
- SimplePDF does not sit between your app and your AI provider
- You choose the model, the keys and where documents live
- The exception: the hosted Copilot demo routes chat through a SimplePDF demo server unless you add your own AI key
Plug in wherever your agent runs
In your app, the browser, the chat or your backend.
AI agent tools in your app
Your model fills the open PDF. Free to try; your own domain needs a plan.Data: the PDF stays in the browser, the model is yours.
Browser agents, through WebMCP
AI agents built into the browser find the editor's tools and fill the form.Data: the PDF stays in the browser, with the user's agent.
Claude, ChatGPT and VS Code, through MCP
List documents, prepare pre-filled forms and read submissions from the chat your team uses.Data: values go to your storage; your chat app sees what the agent reads.
Pre-fill from your backend, through the API
Your backend sends a pre-filled form; the person checks and submits. Needs your own storage.Data: field values go to your storage, never to SimplePDF.
For people who can't fill forms alone
They talk or type to your agent, and a person still checks every answer.
People who can't see the form
Your agent reads each question aloud, fills in the answer, and reads it all back.
People who can't speak
Hard forms usually mean a phone call. Your agent is help they can type to.
Hands that can't hit tiny boxes
People with tremors say the answer; your agent puts it in the right field.
Children who can't write yet
They answer out loud, your agent writes it down, a parent checks and signs.
When reading is the hard part
For dyslexia or low literacy, your agent asks one plain question at a time.
People who don't speak the form's language
They answer in their own language, and your agent writes it in the form's.
Trusted by teams like Centralis Health
Healthcare · HIPAA-compliant workflowsWe needed a customizable PDF viewer that allowed users to modify PDFs upon receipt to fit their specific workflows. SimplePDF’s functionality was precisely what we were looking for, but what truly stands out is the exceptional support and follow-up that Ben and their team consistently deliver.
A fit for your product?
Most of these ring true? Start the free trial.
- Your users fill PDF forms inside your product.
- Your app already has most of the answers.
- You want a person to submit, not a black box.
- You want your own AI provider and your own storage.
What it costs
Try the agent tools and WebMCP free on the default SimplePDF editor.
Pro, $349/month: the agent tools on your own domain, your branding, your S3 or Azure Blob Storage.
Premium, $899/month: the MCP server, the Prefill API, SharePoint and audit logs.
Your AI provider bills you directly for the model; SimplePDF adds no per-token fee.
7-day free trial on every plan.