top of page

The Solution

Our team designed and implemented an AI-driven system with multiple components:

​

  • Vector Database Implementation: Built a PgVector-based database storing merchant profiles as vector embeddings from the MPNet model, enabling semantic search for highly relevant results.
     

  • Chatbot Enhancement: Improved the Rasa-based chatbot by refining intent classes and expanding the dataset linking user messages to intents. This enhanced its ability to respond accurately to merchant search queries.
     

  • Coupon Allocation Mechanism: Leveraged MPNet embeddings to allocate coupons based on user activities, such as check-ins, while factoring in merchant attributes like description and category to ensure personalized offers.
     

  • Content-Based Recommendation System: Developed a system that analyzes user interactions and click behavior to provide tailored recommendations, with future refinements based on implicit feedback to continuously improve relevance.
     

These solutions collectively automated key processes and created a more engaging, personalized platform for users and merchants alike.

Business Impact

The AI-powered enhancements delivered measurable improvements:

​

  • Improved User Experience: Automated coupon allocation and personalized recommendations increased engagement and satisfaction.
     

  • Enhanced AI Capabilities: The upgraded chatbot provided more accurate and meaningful merchant search results, reducing friction for users.
     

  • Streamlined Merchant Interactions: Merchants received better-targeted exposure to users, improving redemption rates and participation.
     

  • Ongoing Innovation: The system supports continuous refinement, ensuring the platform remains adaptive and future-ready for evolving user behavior.
     

These improvements helped the client strengthen community engagement, enhance operational efficiency, and deliver a superior AI-driven experience.

</    Web Development Tools    >

icon-react.png

React

icon-erlang.png

Node.js

Python

icon-pheonix.png

Angular

More projects

OctoPlay

coming soon...

Athena

coming soon...

Backlight

coming soon...

Frequently Asked Questions

How does the AI-powered coupon allocation work?

Coupons are distributed based on user achievements and merchant attributes using vector embeddings to ensure personalized and relevant offers.

How does personalization improve user engagement?

By analyzing user behavior and clicks, the system delivers tailored offers and recommendations that align with each user’s interests and exploration patterns.

Can users find merchants accurately with the chatbot?

Yes, the Rasa-powered semantic search chatbot leverages vector embeddings and refined intent classification to provide precise merchant recommendations.

Is the system scalable for future growth?

The architecture is designed to integrate additional merchants, handle more users, and continuously improve recommendations based on engagement feedback.

Overview

Our client is a community-focused app dedicated to supporting local businesses by helping users discover offerings, check in, and earn discount coupons. By connecting users with restaurants, farmers, and retailers, the platform encourages local economic growth and strengthens community engagement.

​

To enhance the platform’s capabilities, the client wanted to integrate AI-driven functionality, including automated coupon allocation and an interactive chatbot for merchant searches. The goal was to provide users with personalized, relevant experiences while maintaining strong partnerships with local merchants.

Web development stories

Web artefacts

Picto

Mobile app

Picto

Mobile app

Picto

Mobile app

Tech Stack

The Challenge

The client faced several technical and operational challenges:

​

  • Manual Coupon Distribution: Previous coupon allocation relied on manual processes, limiting personalization and scalability.
     

  • Accurate Merchant Search: Users needed an AI-powered chatbot that could provide precise merchant recommendations.
     

  • Integration Complexity: New AI functionalities had to integrate seamlessly with existing systems without disrupting operations.
     

  • User Personalization: The platform required recommendations tailored to individual user behavior and engagement patterns.
     

The business goal was to leverage AI to automate coupon distribution, enhance merchant search, and provide a highly personalized user experience.

AI-Powered Local Business Engagement for a Community App

Brand: Not Disclosed 

Industry: E-Commerce

Technologies: Python, Rasa, Elixir, Phoenix, React Native, Postgres, AWS, Docker, RESTful APIs

Partnership: 2019-2024

bottom of page