Purple8 DocIntel
Any document. Structured knowledge.
Turns every document into the people, money, dates and relationships inside it, ready for your graph.
How DocIntel works
Any document in. A graph out.
Turn documents into entities and relationships with stable IDs. Review the results, then export them to your graph or application. No Hyper Graph installation required.
Hosted content and results stay in memory for the processing session; account, job and billing metadata are retained separately. Choose where your exported knowledge lives. Hyper Graph is optional.
How it is different
Documents become knowledge, not just text
Graph-ready
Built to feed a graph, not a chatbot
Most document tools stop at text chunks for a vector store. DocIntel returns the people, organisations, money, dates and relationships between them, with stable IDs. Download structured results or send them to your chosen graph or application. Hyper Graph is an optional destination, not a requirement.
Two-pass
Entities first, then relationships
A fast entity model finds what is in the document. A language model then links those entities, given the full list as context. Each pass does what it is best at, and every domain profile (contracts, finance, science, AEC) brings its own ontology.
Stable IDs for integration
Normalized names and types produce deterministic IDs for de-duplication. Review ambiguous names before merging records across documents.
Session-based processing
Hosted document content and results stay in RAM for the processing session. Account, job and billing metadata are retained separately. Download results before the session ends.
Native script, every language
Language is detected per document across 75+ languages, and entity names stay in their original script.
Structure-aware chunking
Contracts split on clauses, papers on sections, spreadsheets keep their tables. The split follows the document, not a character count.
Your providers, your choice
Self-host with local OCR and a local language model for in-network processing, or choose cloud providers. Data handling depends on your selected providers.
Ready for your destination
Document type and chunking hints travel with structured output. Use the REST API, export results, or connect an optional graph destination.
Licensing & Deployment
Evaluate in the Cloud Sandbox. License for Cloud, VPC, or Air-Gapped Production.
Test DocIntel immediately in the interactive Cloud Sandbox or connect your AI agent directly over MCP (/mcp). When turning unstructured documents or text into semantic vectors, graph nodes, and relationships in Purple8 Hyper Graph - or running standalone document extraction at enterprise scale - contact us for Commercial or Enterprise licensing across VPC, On-Premise, and Cloud deployments.
How do I start?
$0 Free · Browser & MCP Sandbox
DocIntel Playground ($0)
Evaluate DocIntel immediately in your browser or connect your AI agent over MCP (/mcp, 13 tools).
Try DocIntel$79 · $299 · $599/mo (or Annual)
DocIntel Cloud & Self-Hosted
Start with DocIntel Builder ($79/mo for 10K docs/mo), Production ($299/mo unlimited docs + SharePoint/Confluence/S3 connectors), or Growth ($599/mo multi-tenant + SSO).
View DocIntel Plans →Custom Annual License · Sovereign & Enterprise
Enterprise / BYOC (VPC & Air-Gapped)
Deploy inside your own AWS/Azure/GCP VPC, Red Hat OpenShift (UBI9), on-premise, or air-gapped network with custom domain adapters and 24×7 Severity-1 SLAs.
Contact UsWhat is inside
Formats
- PDF, Word, Excel, PowerPoint, HTML
- Scanned images with OCR
- Email, CAD, BIM, SAP IDocs
Extraction
- People, money, dates, relationships
- Contract, financial, scientific, AEC profiles
- Custom ontologies
Quality
- Stable entity IDs
- De-duplication
- Confidence thresholds per profile
Connectors
- SharePoint
- Confluence
- Amazon S3 and webhooks
Operations
- Async jobs with live progress
- RAM-only hosted document sessions
- Metrics, health and readiness
Output
- Structured exports and REST API
- Preview before sending
- Optional Hyper Graph integration