AI Ticket Routing for Real Estate Operations
- Industry:
- Real Estate
- Role:
- Senior AI Engineer
- Duration:
- 2026 – present
The problem
A property developer operating at scale handles a constant stream of customer tickets — service requests, enquiries, complaints. Each one needs to be understood, worked on, and resolved correctly, with everything logged in the CRM. Done manually, this means queues, re-keyed data, and customers repeating themselves every time a ticket gets passed to someone new.
What I’m building
An AI agent that reads incoming tickets, works out what the customer needs, and works the problem using a knowledge base of company procedures and past resolutions — integrated directly with Salesforce. When the agent can resolve the ticket, it does, and the customer gets their answer. When it can’t — either because the request is outside what it knows, or because something goes wrong partway through — it hands the ticket to a CRE (Customer Relationship Executive) with the full context already attached, so nothing has to be re-explained.
Where it stands
In active development and rollout. Results will be shared when there are numbers worth sharing.
Under the hood
Python, FastAPI, LangGraph, Azure (CosmosDB, Blob Storage), Pinecone for knowledge base retrieval, Salesforce integration.