The offer

From audit to production document AI system.

A clear path to turn sensitive documents into cited answers without losing control of your data, access rights, and evidence.

Deployment

Three deployment models — you choose.

Privacy is a dial, not a switch. Pick the level that fits your constraints — without giving up answer quality.

Recommended

Managed

We run the system end to end. Your documents stay in your private cloud; for each question, only the relevant snippets are sent to a frontier model under an enterprise agreement. Fast to production, no hardware to manage.

Best for most teams that want results fast, without managing infrastructure.

Balanced

Hybrid

Your documents and index stay with you. Reasoning and tool calls use frontier models via enterprise API — no training on your data, zero-retention available. The best quality-to-privacy trade-off.

Best for regulated teams that must keep documents in-house without giving up quality.

Maximum sovereignty

On-premise

Everything runs inside your perimeter, models included, on your infrastructure. For the strictest regulatory constraints. Custom quote based on the hardware required.

Best for defense, classified data, or strict data-sovereignty requirements.

Your documents stay inside your perimeter — only the needed snippets are queried.
Frontier models under enterprise agreement: no training on your data, zero-retention available.
Built to support GDPR and EU data residency.
Initial phase

Private RAG Audit & Architecture

?3,000 - ?5,000

Validate the use case, measure risks, and define the document architecture before building.

  • Audit of your sources: PDF, Word, emails, reports, contracts
  • Mapping of volumes, formats, access rights, and risks
  • Target architecture: server, models, vector database, interface
  • Prioritized deployment plan with first use case
  • Deliverable report to support the internal project decision
Main delivery

Document AI System Build

?8,000 - ?15,000

A private RAG system connected to your documents and usable by your teams.

  • Ingestion and indexing pipeline on your infrastructure
  • Private vector database adapted to your access rights
  • AI models configured to your business vocabulary
  • Natural language search interface with cited answers
  • Quality tests on your real questions and sensitive cases
  • Team training and full technical documentation
Main delivery
Monthly

Maintenance & Evolution

?2,000 - ?3,500 / month

The system stays useful as your files, teams, and risks evolve.

  • Monthly ingestion of new files and corpora
  • Pipeline, prompt, and citation rule updates
  • Quality monitoring: retrieved sources, weak answers, errors
  • Priority support with response within 24 business hours
  • Monthly usage, risk, and recommendation report

Prices are excluding taxes. The exact scope is confirmed after the initial audit based on document volumes, hosting constraints, and expected governance level.