Cutting 500+ Daily Mall Queries from Hours to Seconds

The Challenge

The mall's info desks were fielding 500+ queries a day across 200+ tenants. During peak hours, the team couldn't keep pace with demand. They needed a way to scale service quality without growing headcount.

What We Built

We built an AI agent that pulls in every tenant's store and product data, organizes it, and fields shopper questions without adding to the team's workload. It combines a language model with a retrieval layer, a recommendation engine tuned to visitor preferences, and a memory system that carries context from past conversations. The whole thing responds in under 2 seconds, as fast as a live concierge.

Results

500+ daily queries now resolve automatically. Store recommendation engagement doubled. Response accuracy reached 9 in 10. The system offset 3 full time salaries in Q1, covering the build cost within 90 days.