Introducing ParseApi: Turn Any Folder of Documents Into a Live API
Every developer has been in this situation: a client hands over a folder of PDFs and says "we need the data from these." What follows is a multi-day deep dive into PDF parsing libraries, OCR toolchains, schema design, normalization scripts, and REST API boilerplate — all before you've answered the actual question.
I've built that pipeline four times. For invoices. For medical records. For legal contracts. For resumes. Each time with slightly different tooling, slightly different schemas, and the same fundamental plumbing underneath.
ParseApi is the version of that pipeline I kept building from scratch every time.
What it does
You create a folder. You drag in documents. You get a live REST API endpoint that returns structured JSON.
GET https://api.parseapi.dev/v1/yonas/invoices
The response is a paginated list of extracted data. Every document you add gets processed automatically. The schema is detected from your first few uploads. You can edit individual fields if the AI gets something wrong. The API is live the moment you create the folder.
The folder is the unit
Most document extraction tools are document-centric: you upload a file, you get back JSON. ParseApi is folder-centric, and that changes everything.
A folder has a slug. That slug becomes your API endpoint. The folder accumulates documents over time. When you add a new invoice, it joins the same endpoint as the previous ones. You can query the entire dataset, filter by field values, or pull the latest document.
This means the folder is your integration point. Your accounts-payable team uploads invoices every Monday. Your ERP system queries the same endpoint every Tuesday. Nobody writes glue code.
Auto-schema detection
When you upload your first three documents to a new folder, ParseApi runs them through an open-ended extraction pass and synthesizes a schema from the common fields. You see a field list — invoice_number, vendor_name, line_items, total_amount, due_date — and you can rename, remove, or add fields before it applies to future uploads.
Once the schema is confirmed, every subsequent document conforms to it. If a field is missing from a particular document, the value is null rather than absent.
Editable extractions
AI gets things wrong. A vendor name is truncated. A date is in the wrong format. A line-item amount is misread.
ParseApi doesn't hide these errors — it surfaces them. The extraction view shows your document and the extracted data side by side. You click a field to see where in the document it was found. You edit it inline. The correction is recorded and the API immediately reflects the new value.
Over time, corrections become a feedback signal for understanding where extraction accuracy is lowest.
What's live today
- Upload via the web dashboard or the REST API
- Auto-schema detection after the first three uploads
- JSON API with cursor-based pagination
- API key authentication (with JWT, Basic, and public modes also available)
- Public or private folders — per folder, never platform-wide
- Editable extractions with source highlighting
- Admin panel for configuring AI providers and routing rules
What's coming
The roadmap for the next few months includes webhooks (fire-and-forget to your endpoint on every extraction), real-time status updates via SignalR, folder-level search, a CLI tool for local upload workflows, and a library of pre-built schema templates for common document types.
Get started
ParseApi is free for 100 pages per month. Sign in with Google or GitHub, pick a username — it becomes your permanent API namespace — and create your first folder.
Start free at app.parseapi.dev
No credit card required.