Document infrastructure
for the AI era.
Professional documents without the manual work. Describe what you need, your AI agent delivers a publication-quality PDF.
curl -X POST https://api.pressa.dev/api/v1/compile \
-H "Content-Type: application/json" \
-d '{"latex":"\\documentclass{article}\\begin{document}Hello\\end{document}"}'Real typesetting, not HTML hacks
HTML-to-PDF tools were never designed for professional output. Pressa compiles with LaTeX - the same engine behind every academic journal and technical publication.
Looks like it was made in 2005
System fonts, zero attention to typography
Broken layout and spacing
Margins overflow, tables break across pages
No math or formula support
Equations render as broken images or plain text
Publication-ready from the first compile
Professional fonts, precise kerning and spacing
Pixel-perfect layout on every page
Tables, headers, footers - all exactly where they should be
Native math, formulas, and symbols
Publication-quality typesetting - not HTML hacks
Two commands, then it just works
Install once. Every AI tool on your machine gains document generation from that point on.
Install Pressa
Run two commands. Pressa registers as a tool your AI assistant can use. No configuration needed - it just works.
$ npm install -g pressa
$ pressa auth --key YOUR_API_KEY
# That's it. Your AI can now create PDFs.Describe your document
Tell your AI what you need in plain language. Be as specific or as vague as you want - it figures out the rest.
"Create a quarterly sales report with charts for Q1 2026. Include revenue by region, top 10 accounts, and a comparison to last quarter."
Iterate until perfect
Not quite right? Just say what to change. Your AI updates the document and gives you a new PDF instantly.
"Add an executive summary on page 1 and move the charts to an appendix. Also bump the font size up a notch."
Drop into any stack
Your backend sends data, an LLM describes the document, Pressa compiles the PDF. One endpoint, any language, production-ready output.
How your app talks to Pressa
Your app sends document data to the Pressa API and gets a PDF URL back. You describe what you need - an LLM handles the formatting for you.
The complete integration
Three steps in your backend: get your data, describe the document, send it to Pressa.
// 1. Get data, 2. Describe document, 3. Get PDF
const invoice = await db.invoices.find(id);
const latex = await llm.generate(
`Create an invoice for ${invoice.company}...`
);
const pdf = await fetch(
"https://api.pressa.dev/api/v1/compile",
{ method: "POST", body: JSON.stringify({ latex }) }
);
// pdf.pdf_url → "https://api.pressa.dev/api/v1/pdfs/a1b2c3"
// pdf.compilation_time_ms → 842Works with any language. Full API reference in the docs.
No key needed to start. A free account raises the limit to 50 compilations per month and adds image assets, saved templates and the render API.
Iterate without re-prompting
The first AI compile is rarely the final one. Pressa Studio opens any rendered PDF in your browser, maps every click to the LaTeX line that produced it, and lets you fix the result directly. Surgical edits, not "make the title bigger" prompts.
Click a pixel, see the line
Built on SyncTeX. Click anywhere on the rendered PDF and the sidebar shows the exact LaTeX line that produced it. Sub-50ms resolve, no extra setup.
Two ways to fix a line
Seven deterministic helpers (Bold, Italic, Bigger, Smaller, V-Space, Pad Left, Pad Right) for the everyday tweaks. Plain-English prompt to Claude for everything else. Every AI edit is security-validated before it lands.
Click to insert placeholders
Open any saved template in template mode and click a region in the preview to mark it variable. The clicked text becomes a placeholder. The Render API fills it on every future call.
Available on Starter, Pro, and Business plans. Free tier unlocks click-to-locate.
Built for this
Not a PDF library with an API bolted on. Purpose-built infrastructure for AI-driven document generation.
Sub-second compilation
Documents compile in 800ms on average. Fast enough to generate inline - your users never wait.
Every AI, one integration
Claude, ChatGPT, Cursor, Copilot. Install once - every AI tool on your machine gains document generation.
Secure by design
Isolated workspace per compilation with no-shell-escape. Signed URLs that auto-expire. Source deleted immediately, PDFs gone in 24 hours. Built for sensitive documents from day one.
One endpoint, any language
POST your source, GET your PDF. Three compilers (pdflatex, xelatex, lualatex), any language, any framework. Integration takes minutes.
Predictable pricing
Start free. Pay only for what you compile.
Starter
500 compilations / month
- ✓Up to 20 pages / document
- ✓10 saved templates
- ✓Email support
Pro
2,000 compilations / month
- ✓Up to 100 pages / document
- ✓All compilers incl. lualatex
- ✓50 templates & priority queue
Business
10,000 compilations / month
- ✓Up to 500 pages / document
- ✓Unlimited saved templates
- ✓Priority support & 99.9% SLA
Annual plans available - save up to 22%
Compare all plans →Frequently asked questions
Everything you need to know.
Pressa works with any AI assistant that supports MCP (Model Context Protocol) or CLI tools - including Claude, ChatGPT, Cursor, Windsurf, and GitHub Copilot. Install once, and every AI tool on your machine can create professional PDFs.
Not at all. You describe what you want in plain language, and your AI assistant handles the rest. Pressa uses LaTeX under the hood for professional typesetting, but you never need to see or write any of it.
Yes. Every document is compiled in an isolated workspace with no-shell-escape enabled. Source is deleted immediately after compilation. PDFs are delivered via signed URLs that auto-expire after 24 hours. Nothing is stored permanently.
You get a detailed error response with the compilation log and the specific line that caused the issue. Your AI assistant can read this error and automatically fix it - no manual debugging needed.
Anything that needs professional formatting: invoices, reports, contracts, academic papers, resumes, certificates, and more. Pressa uses the full TeXLive distribution under the hood with three compilers and thousands of packages. pdflatex and xelatex are available on every plan; lualatex is included on Pro and Business.
One endpoint. POST your document source to the API, get a signed PDF URL back. Average response time is under one second. SDKs for Ruby, Python, and Node are coming soon, but the REST API works with any language today.
The CLI tool and MCP server are open source (MIT license) and available on GitHub. The compilation API is a hosted service with a generous free tier.
Add document generation to your stack
One API. Any AI. Publication-quality output. Free to start, no credit card required.