100% on-premise · AI runs on your servers

Intelligent document management. Private by architecture.

Doc Lexora classifies, organizes and understands your business documents with AI that runs entirely on your own infrastructure. No cloud APIs. No per-user fees. Your documents never leave your premises.

Self-hosted · Runs as managed system services on your servers · Built on proven open-source technology
100+REST API endpoints
3search modes — full-text, semantic, conversational
5secrecy levels with user clearance
0documents sent to external clouds
AI capabilities

AI that works for you — on your hardware

Doc Lexora integrates a local large language model through Ollama. Classification, search and document analysis run on servers you control — the intelligence of modern AI tools, without the cloud.

Automatic document classification

Every document that enters the system — uploaded, scanned from a watched inbox folder, or imported in bulk — is read and understood. The AI suggests title, document type, correspondent, tags, date and classification code.

  • Confidence score on every suggestion
  • High-confidence results applied automatically, the rest queued for human review
  • Adaptive learning: the system improves from every approval and correction
93% 85% 58%

Search that understands meaning

Three layers of search work together on the same archive:

  • Full-text — exact words and phrases, with filters for type, tags, dates, correspondents and secrecy level
  • Semantic — vector embeddings (pgvector) find documents by concept, not just keywords
  • Conversational (RAG) — ask questions in natural language and get direct answers with source citations, powered by a LightRAG knowledge graph

Deep document understanding

Doc Lexora parses documents with IBM Docling: headings, paragraphs, tables and images are extracted with their structure across PDF, Word, PowerPoint, Excel and HTML. Scanned documents go through multilingual OCR (Tesseract or local vision models).

  • Structured extraction feeds the AI classification and the knowledge graph
  • Intelligent chunking keeps tables and context intact for retrieval
  • OCR in English, Italian and Finnish — including compound words
Private by architecture

Your documents never leave your servers

This is the core architectural decision that defines Doc Lexora: every AI operation — language model, embeddings, vision OCR — runs on hardware you own. Document content is never transmitted to any third-party cloud service.

GDPR compliance by design

Personal data is never processed by external services. No data processing agreements with AI providers, no cross-border transfers, full data sovereignty and a complete audit trail of every access.

Smaller attack surface

No external API calls, no risk of confidential content reaching third-party training data. Five-tier secrecy classification with per-user clearance keeps sensitive documents inside your security perimeter.

Operational control

AI availability doesn't depend on a vendor's uptime. No per-query API costs that grow with usage, no vendor lock-in: model selection and configuration are fully in your hands.

Enterprise features, built in

Everything professional document management requires

Security & access control

Two-factor authentication (TOTP), customizable roles and permission matrices, five secrecy levels with user clearance, API keys for integrations, account lockout and IP logging.

Document lifecycle

Version history with comments and checksums, digital signatures (RSA, PDF embedding), retention policies with automatic enforcement, soft-delete with recovery, hierarchical cabinets with multi-filing.

Workflow automation

State-machine workflow engine with configurable transitions, auto-trigger rules based on document properties, structured input fields at each step, full transition logging.

Background processing

Reliable job queues for OCR, AI metadata, embeddings and knowledge-graph indexing. Watched inbox folders: drop files in, Doc Lexora processes them automatically. Scheduled maintenance built in.

Truly multilingual

Full interface in English, Italian and Finnish with locale-aware routing and dates. Multilingual OCR and AI metadata extraction across all three languages.

Auditability & API

Complete audit trail of every operation, dashboards with document statistics and activity, backup and restore from the admin panel, and a REST API with 100+ endpoints for integrations.

Why Doc Lexora

Enterprise capabilities, without the cloud compromise

Traditional enterprise DMS platforms rely on cloud-connected AI: your document content is processed on external infrastructure, at per-user prices that grow every year. Doc Lexora delivers comparable capabilities on your own servers.

Capability Doc Lexora Traditional cloud DMS
AI document classificationLocal LLMCloud AI
Conversational search with citations (RAG)
Semantic vector searchRare
Knowledge graph across the archive
Workflows, versioning, audit trail, signatures
100% on-premise dataOptional
100% local AI processing
Per-user monthly licence feesNone€40–90 /user/month
Who it serves

Built for organizations that take data seriously

Law firms

Client confidentiality is non-negotiable. Secrecy classification, audit trails and local AI keep case files under your control — while conversational search saves hours of manual review.

Accounting firms

High document volume meets automatic classification: invoices, contracts and reports are tagged and filed without manual effort. Retention policies keep you compliant.

Healthcare & public sector

The strictest GDPR requirements, five-tier classification, complete audit trails and data sovereignty — met by architecture, not by policy.

Any data-sensitive business

If you handle contracts, personnel files, financial records or client data, you get AI-powered document management without giving up privacy.

Architecture

A production-grade architecture, on your infrastructure

Doc Lexora installs on a dedicated server or virtual machine through a reproducible provisioning process. The application and its processing workers run as managed system services with automatic startup and recovery; containerized deployment is available as an alternative. Health-check endpoints, scheduled backups with rotation and automated maintenance are built in. Recommended hardware: 16+ GB RAM for AI workloads; GPU optional but recommended.

Doc Lexora — Web interface + REST API + DMS engineNext.js · React
PostgreSQL + pgvectormetadata · full-text · embeddings
Redis + BullMQbackground job queues
Ollamalocal LLM · embeddings · vision OCR
⬢ All services run on your infrastructure — no external connections, no cloud dependencies.

See Doc Lexora in action

Request a personalized demo or a trial deployment on your own infrastructure. We'll show you what local AI can do with your documents — without your documents ever leaving your premises.

Request a demo