Listening to what customers actually say
I asked our AI agent to make a first pass through our Google reviews and the public feedback on Cindy's Facebook Page. The patterns were useful, but the sample has clear limits.
Since opening Chloe's Mama Kitchen, customers have left their thoughts in different places. Cindy has her Facebook Page, and people also review the shop on Google. I wanted to see whether an AI agent could help me read those scattered comments as one small piece of customer research.
On 23 September, I asked the agent to try it. It read the ten customer reviews visible on our Google Maps listing, the recommendations visible on our Facebook Page, and a sample of recent public posts. This was a first, read-only pass. It did not go through private messages or our sales records, and it did not reply to anyone.
Laksa stood out in the Google reviews: Google's review topics marked it in four of the ten. Customers also described the curry puffs, kuih, chicken wings and Bubur Cha Cha. On Facebook, some recommendations gave more detail about flaky curry-puff pastry, the texture of Angku Kuih and a sweetness level those customers liked.
I found the less flattering detail useful too. One Google reviewer loved the food but said the older shopping-centre setting was plain; that review gave the atmosphere a lower score than the food and service. It is one person's experience, not a trend, but it gives us something specific to think about when we explain the shop and the pickup experience.
The agent had to separate different kinds of evidence. Facebook's recommendation total includes feedback from years before the shop opened. Recent menu posts that we sampled had few substantive public comments. A mention of laksa in a review is not proof that it outsells other dishes, and a customer's ingredient or health impression is not a claim we should repeat as a verified fact.
This first pass gave me a short list of questions: which dishes do customers describe in their own words, where does the experience surprise them, and what could we explain more clearly? We have not changed the menu or the shop based on this pass. For now, the useful result is a more careful way to listen before deciding what to do next.
