Ifeoluwa Samuel
Back to work

Case Study 07

AI Customer Support Chatbot with RAG (Real Estate)

A Telegram-based AI chatbot that retrieves property listings from a live knowledge base and answers prospect enquiries in real time, qualifying leads and alerting agents automatically.

  • n8n
  • Telegram Bot API
  • Groq AI
  • LLaMA 3.3-70b
  • Google Sheets
  • Airtable
  • Gmail
  • Slack

The business problem

Agents answered the same questions all day: price, location, number of bedrooms, availability, and viewing times. Enquiries came in at every hour, including evenings and weekends, and buyers who did not get a quick reply moved on to the next agency. Answers also varied between agents, and details taken from memory were sometimes out of date because the listing sheet had already changed.

The solution

An n8n workflow is triggered by each Telegram message. It reads the current property listings from Google Sheets and injects the relevant rows into a Groq AI system prompt, so the model answers strictly from live inventory rather than from training data and cannot invent a property that is not for sale. The reply goes straight back to the prospect in the chat, usually within a couple of seconds. When the conversation shows real buying intent, the workflow writes the lead and the full transcript to Airtable and alerts the assigned agent by email and Slack, so a human takes over at exactly the right moment.

Walkthroughs for this build

AI Customer Support Chatbot with RAG (Real Estate) workflowAI Customer Support Chatbot with RAG (Real Estate) workflowAI Customer Support Chatbot with RAG (Real Estate) workflowAI Customer Support Chatbot with RAG (Real Estate) workflowAI Customer Support Chatbot with RAG (Real Estate) workflow

Want a system like this in your business?

Every build starts with a short conversation about where the manual work actually lives.

Book a Discovery Call
Next projectAutomated Lead Capture & CRM System