Overview
We built an end-to-end AI assistant that helps Gibraltar estate agents surface the right listings faster, respond to tenant or buyer requests in natural language, and keep inventory synced without manual data entry. The platform unifies crawling, enrichment, and a mobile-ready experience into a single tool the sales floor uses every day.
Client Background
The agency is a mid-sized residential lettings and sales team with storefront offices serving the Gibraltar market. The team juggles a large pipeline of expatriate clients, each with tight relocation timelines. Their agents rely on timely, accurate inventory data and quick answers about amenities such as pet policies to close deals.
The Challenge
Agents were spending hours reconciling listings from disparate portals, copying details into spreadsheets, and chasing down answers to common buyer questions. Manual workflows delayed replies to inbound leads, and data gaps (pet policies, pricing updates, duplicate records) were damaging client confidence. Leadership needed a system that could ingest new stock automatically, normalize fields, and let agents ask conversational questions like “two-bed apartments with pet-friendly terms under £2,500 in Ocean Village.”
The Solution
We delivered a vertically integrated platform spanning acquisition, enrichment, and agent-facing experiences:
- A Scrapy-based crawler and Firecrawl ingestion pipeline deduplicate and upsert listings directly into Supabase/PostgreSQL via hashed fingerprints.
- A Node/Express API backed by Prisma exposes listings, filters, and detailed records, powering both web and mobile (Capacitor) clients.
- An OpenAI-powered service parses free-text buyer prompts into structured filters and classifies pet policies at scale, keeping critical amenity data fresh.
- A React + Tailwind UI with table/grid toggles, saved searches, and deep links equips agents to collaborate around shared criteria in real time.
Key Features:
- Natural-language query inbox that converts client requests into actionable listing filters in seconds
- Automated ingestion and deduplication from partner portals with Supabase-backed persistence
- AI enrichment of features including pet-allowance and attribute gaps to boost buyer confidence
- Mobile-ready React experience delivered via Capacitor for on-the-go viewings
Tech Stack:
React, Vite, Tailwind CSS, Express.js, Prisma,
PostgreSQL (Supabase), OpenAI, Scrapy
Results
Within the first month of rollout the agency’s floor team replaced spreadsheet juggling with guided, AI-assisted workflows.
Highlights:
- ⏱️ Saved 12+ agent-hours per week previously spent on manual data entry and reconciliation
- 🚀 Cut time-to-reply on buyer requests from ~4 hours to under 15 minutes using conversational search
- 💰 Lifted qualified viewing bookings by 22% thanks to richer amenity tagging and faster follow-up
- 📦 Keeps a rolling catalogue of 500+ active listings synced nightly without duplicate records
Key Takeaways
Combining reliable data ingestion with AI-powered interpretation unlocked a faster, more confident sales process. Supabase and Prisma keep the stack maintainable, while React + Capacitor ensure agents can work the same way at a desk or during property tours. Next steps include expanding the crawler catalog and layering in buyer-matching automations for inbound leads.