Document Workbench
Prototype prompts, schemas, OCR settings, source locations, tracing, and map-reduce in one authenticated workspace.
AI DOCUMENT OPS
DocuDevs.ai
Production document processing for teams that need data, not PDFs.
Extract structured JSON from documents, prove where every field came from, and automate the whole intake workflow with SDKs, jobs, batches, agents, and the new visual integration builder.
Source locations Visual flows Python + Java SDKs BYO LLM/OCR
Any file PDF, Office, scans, images, spreadsheets
Any shape Prompt, JSON schema, templates, calculated fields
Any workflow API calls, batches, cases, agents, edge flows
Any stack Python, Java, cURL, webhooks, BYO providers
CONTROL PLANE
One platform from first extraction to live workflow.
DocuDevs combines a developer API with an operations UI, so teams can prototype, inspect, automate, and govern document processing without rebuilding the same plumbing.
Integration Builder
Build visual flows for document intake, OCR, extraction, routing, and exports, then keep the generated YAML for review.
Edge runtimes
Run embedded integrations near private files while the cloud control plane manages configs, status, and orchestration.
Batch operations
Process folders of documents, track progress, and download normalized JSON, CSV, or Excel outputs when the run completes.
Agent assisted setup
Describe the outcome you need and let the agent help draft extraction instructions, schema fields, and processing plans.
Provider control
Use DocuDevs managed OCR and LLMs, or connect your own Azure Document Intelligence, OpenAI, and Azure OpenAI providers.
NEW: INTEGRATION BUILDER
Drag together document workflows, keep the YAML.
The builder turns common integration work into a visual canvas: watch a folder, read documents, extract with a saved configuration, branch on values, and ship results to the next system.
- Visual canvas for flow design
- Account-wide or runtime-specific configs
- Edge runtime handoff for private environments
FLOW BUILDER ready
Visual canvas drag-drop
Generated config YAML
Runtime scope account or edge
tasks: - watch: /invoices - ocr: layout - extract: invoice_v3 - post: ERP_WEBHOOK_URL
DEVELOPER SURFACE
Use the SDK when code is right. Use the builder when workflow is right.
The same extraction engine powers API requests, UI jobs, saved configurations, batch runs, and edge flows.
Typed extraction
from docudevs import DocuDevsClient
from pydantic import BaseModel
client = DocuDevsClient(token="your-api-key")
class Shipment(BaseModel):
container_id: str
port_of_loading: str
gross_weight_kg: float
result = client.extract(
"bill-of-lading.pdf",
schema=Shipment,
source_locations=True
)Validated object
Shipment( container_id="MSCU4829910", port_of_loading="Helsinki", gross_weight_kg=21740.0 )
DOCUMENT PIPELINE
A clean path through messy inputs.
Every document run is observable: queued, processed, traced, stored, and ready for the next system.
Describe the data
Start from a prompt, JSON schema, saved configuration, template, or agent drafted extraction plan.
Read the document
OCR converts layout, tables, figures, spreadsheets, and long reports into model-ready context.
Extract and verify
Structured output is validated, enriched with source evidence, and tracked as an async job.
Ship the result
Return JSON to your API, export CSV or Excel, fill templates, or push through an integration flow.
USE CASES
Built for the documents that slow teams down.
Start with a single extraction and grow into cases, templates, quality checks, batch exports, and integration flows.
Purchase orders and invoices
Contracts and legal reviews
Claims and case files
Shipping and customs documents
Financial reports and tables
Compliance packets and audit evidence
START FAST
Pick the entry point that matches your team.
READY
Turn the next document workflow into a production system.
Start with the full platform, a token pack, and the same API surface you can keep in production.