Loading AI template builder
Use AI to draft templates from prompts, screenshots, reference PDFs, HTML, or sample JSON. Map fields visually. Publish a reusable template. Call it from your API.
Start from a prompt, screenshot, PDF, HTML snippet, or sample JSON. The builder turns the reference into a template draft your team can review.
Connect JSON paths like customer.name, line_items[], totals.amount, and due_date to specific document regions.
Render sample data, check layout issues, test long fields, and confirm that mapped values land where they should.
Approved templates become available in the workspace you choose, with a stable templateId developers can call through the API.
Bring the document you need: a prompt, a screenshot, an old PDF, a Figma export, HTML, or a rough sketch.
Use fake representative data. Do not submit PHI, secrets, or sensitive customer data while drafting templates.
Point each section of the document to the right object, array, field, total, date, or conditional value.
Once reviewed, publish it as a reusable API template and generate documents through POST /v1/generate.
Own the look of your documents without filing an engineering ticket. Upload a screenshot, describe what you need, click publish.
Non-technical teams own the design. You get a stable templateId to call through the API. No redeploys when the design changes.
Standard PDF generation uses published templates and structured payloads without requiring a model call. Builder inputs may be sent to an AI provider as described in the Privacy Policy and sub-processor list.
Use fake representative data when drafting templates. Do not submit PHI, secrets, or sensitive customer data during drafting.
Once your team publishes a custom template, developers call it exactly the same way they'd call a built-in. Same endpoint, same auth, stable templateId.
{
"templateId": "custom_invoice_v1",
"data": {
"customer": {
"name": "Acme Studio"
},
"line_items": [
{ "description": "Design retainer", "quantity": 1, "unit_price": 2450 }
],
"totals": {
"amount": 2450
}
}
}A quick pass to catch the issues that bite in production.