étude de casReal Estate

Real Estate Company: RAG Knowledge Base for €1.2B in Assets

Confidential real estate company
20 septembre 2024
Résultats Clés
-60% search time
15k+ documents indexed
98% response accuracy

How a real estate company managing €1.2 billion in assets reduced document search time by 60% with an internal RAG chatbot.

Context

Client name anonymized for confidentiality reasons.

This French real estate company manages a portfolio of over €1.2 billion in assets: offices, retail, logistics. Their team of 45 people (asset managers, lawyers, property managers) handles thousands of documents daily:

  • Commercial leases and amendments
  • Management contracts and mandates
  • Urban planning regulations
  • Expert reports and audits
  • Tenant correspondence

The problem: an asset manager spent an average of 2h30 per day searching for information in these documents scattered between servers, emails and digitized physical archives.

The solution

We deployed an on-premise RAG knowledge base accessible via an internal web interface and integrated with their Azure AD SSO.

Key features

  1. Natural language search

    "What are the renewal clauses for the Carrefour lease on the Lyon Part-Dieu asset?"

  2. Automatic data extraction

    • Lease expiration dates
    • Rent amounts and indexations
    • Special clauses (works, termination, etc.)
  3. Proactive alerts

    • Notification 6 months before lease expiry
    • Detection of unusual clauses
  4. Complete audit trail

    • Access traceability
    • Sources cited for each response

Technical architecture

  • Deployment: On-premise on client infrastructure (no data in the cloud)
  • Ingestion: SharePoint connectors, emails, OCR scanner
  • Security: Azure AD SSO, access logs, AES-256 encryption
  • Model: Self-hosted LLM (Mistral 7B) for GDPR compliance

Results

After 6 months of deployment across all teams:

MetricBeforeAfterChange
Search time/day2h301h-60%
Data entry errors12/month2/month-83%
Tenant response time48h12h-75%

Estimated ROI: €320k/year in time saved (45 people × 1h30/day × 220 days).

Testimonial

"Our asset managers have regained time for analysis and client relations. Document search had become a bottleneck, it's now a competitive advantage."

Director of Operations

Tech stack

  • Python, FastAPI
  • PostgreSQL, Qdrant
  • Mistral 7B (on-premise)
  • Azure AD SSO
  • Docker, Kubernetes

Project completed in 12 weeks. Contact us for an on-premise deployment.

Tags

real estateRAGknowledge baseenterpriseon-premise

Prêt à transformer votre entreprise ?

Déployez un chatbot intelligent en 5 minutes avec Ailog RAG as a Service.

Essayer gratuitement

Articles connexes

Ailog Assistant

Ici pour vous aider

Salut ! Pose-moi des questions sur Ailog et comment intégrer votre RAG dans vos projets !