Quickstart Guide
Get started with Sea.dev's document intelligence APIs.
Extract structured data from documents with full citation tracking, enabling auditable and transparent AI-powered workflows.
Key Concepts
Four concepts cover everything you need to get started. For the full set of objects, including Chat, and the distinction between extraction and chat, see Core concepts.
- Deals are the container for one client relationship or transaction. Every document, extraction, and result links back to a deal.
- Forms define the structured fields you want to extract from a document — the target shape of your data.
- Submissions hold the data extracted for a form from a specific document, complete with citations and a full audit trail.
- Pipelines are a configurable sequence of steps that process a deal’s documents end to end — extracting data, cross-checking it, calling out to your systems, and routing low-confidence items to a reviewer. Each time a pipeline processes a deal it creates a pipeline run.
Data model at a glance
Deal
The container for one client or transaction. Everything links back to a deal.
↓
→
Pipeline
Runs a sequence of steps over the deal’s documents: extract, cross-check, integrate, and review.
↓
→
Form
Defines the structured fields each step should extract.
↓
→
Submission
The extracted values for a form, with citations and a full audit trail.
Extract Data via App
- Sign up at app.sea.dev
- Create a "New Submission" for any existing form
- Upload a PDF document, and it will be extracted into structured data
- Check the citations, edit the form, and re-run the submission as you require!
- Once you've created your first submission, feel free to set up a new Deal, Form, or Pipeline.
Use the API
- Navigate to Settings > API Keys
- Click "New API Key"
- Copy and securely store your API key
- Create a submission via API at api.sea.dev/docs, grab a Deal ID and Form ID from the frontend.
Keep your API key secure. Never commit it to version control or share it publicly.