Reply Desk: a review-reply assistant for small shops, where customers' words never leave the counter
This is a submission for the Hacktoberfest Open-Source AI Challenge: Week 2 — Side Hustle . What I Built Reply Desk is a review-reply assistant for small businesses: the chai stall, the salon, the neighbourhood clinic, the coaching centre, the home baker. These businesses live and die on Google and Zomato reviews. Replying to every review helps, and nobody behind the counter has time to do it.…
This article is about Reply Desk, an open-source AI tool designed to help small businesses manage customer reviews and provide timely responses. The software, created by Yash11778, uses the Gemma 3 1B open-weight model to sort reviews based on urgency and mood, then drafts replies in the business owner's voice.
The tool flags reviews related to food safety, health, money, and staff behavior, moving them to the top of the list. It also identifies the mood of each review – positive, mixed, or negative – and adjusts the response accordingly. For negative reviews, the assistant suggests contacting the owner directly via a provided phone number or email, rather than engaging in potentially hostile public discourse.
Users simply paste their Google or Zomato reviews into the tool, which then processes them to generate replies. The system runs entirely in the browser, requiring no server, API key, or account. The code is open-source and available on GitHub, with a MIT license.
According to the developer, the tool is aimed at small businesses that rely on reviews for their success, such as cafés, salons, clinics, and home bakeries. The pricing model allows businesses to answer every review within two minutes for a monthly fee of ₹199–299.
The author tested the tool with a fictional café called Chai Point Corner, run by the character Priya. The generated replies were polite, specific, and signed by the owner, demonstrating the tool's ability to provide professional responses without inventing false information. The system was able to handle the task quickly, even on a CPU-only laptop, with the first run downloading the model (~0.8–1 GB) and subsequent runs taking only a few seconds per reply.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
