RAG-Powered Property Management: How IgeraFincas Answers in Seconds
The problem with generic AI chatbots
ChatGPT doesn't know your community bylaws. Standard chatbots don't know that the AGM quorum in your building is 40%, or that parking bays 12–18 are subject to a special restriction in your lease. IgeraFincas does — because it reads your actual documents.
Definition
What is Retrieval-Augmented Generation (RAG)?
RAG is an AI architecture that combines a large language model (LLM) with a knowledge retrieval system. Instead of answering purely from training data, the AI first retrieves relevant excerpts from your uploaded documents, then generates an answer grounded in those specific passages. Every answer cites the exact source paragraph — no hallucinations.
IgeraFincas' 3-layer knowledge architecture
Layer 1 — Public Law (pre-loaded)
Spanish LPH (Ley 49/1960), Catalan Codi Civil (Art. 553), UK Leasehold and Freehold Reform Act 2024, Florida HOA Act §720, California Davis-Stirling Act. Always current. Zero upload required.
Layer 2 — Community documents (per building)
Bylaws, CC&Rs, estatutos, reglamento interno, minutes of AGMs. Uploaded by the property manager. Chunked, embedded with Gemini Embedding 2, and stored with pgvector for sub-200ms semantic search.
Layer 3 — Operational documents (per building)
Maintenance contracts, insurance policies, supplier agreements, outstanding invoices. Gives context for service charge disputes, contract renewal dates, and insurance coverage questions.
How it works in practice
❓ "What's the quorum for our AGM to be valid?"
✅ "According to your community's bylaws (Article 7.3, uploaded 14 May 2026), the AGM requires a quorum of 40% of ownership shares in first call, and can proceed on second call regardless of attendance after 30 minutes. The LPH Art. 16.1 sets the minimum at one-third as a legal floor — your community has set a stricter threshold."
⏱ 180ms · 📄 Community Bylaws Art.7.3 + LPH Art.16.1 · 🚫 0 hallucinations