Effectiveness of Digital Activities with AI

Verysell AI supported the development of an AI-driven digital platform for a food and beverage company in East Asia, with a strong focus on enhancing system integration, user experience, and security. The project aimed to leverage AI technologies to streamline user authentication, notification services, and data management within the app. SmartDev provided full-stack development, AI-powered features, and rigorous quality assurance to deliver a secure, intuitive, and intelligent user experience. Key AI-driven components included Firebase-Supabase authentication enhanced with machine learning models for user verification, real-time notifications powered by predictive algorithms, and advanced session management using AI to improve security and user interactions.

* The client is under NDA with Verysell AI

Headcounts

1 Tech Lead, 2 Developers, 1 Mobile Developer, 1 QA Engineer

Industry

Food and Beverage

Products and Services

Java, AI, ReactJS developer, Tester

Timescales

Ongoing (Since 2025)

Country

Japan

The Challenge

The client sought to disrupt the beverage industry with a market-first, the client faces various challenges that affect their profitability:

  • Integration Issues: Difficulty in integrating third-party authentication systems, such as Firebase Auth, with Supabase.
  • UI/UX Bugs: Multiple UI bugs impacting user experience, including problems with session reordering, profile viewing, and form data retention.
  • Notification and Email Services: Challenges in selecting and integrating appropriate email and push notification services.
  • Security and Data Management: Ensuring the secure handling of personal data and the implementation of robust security protocols.

Technology Stacks

Java

Flutter

React Native

The Solution

Verysell AI delivers various AI solutions:

  • Integration Solutions: We integrated Firebase Auth with Supabase to manage secure access to AI services. JWT tokens and Row-Level Security (RLS) policies were implemented to enforce granular data access, protecting AI model inputs and outputs. Custom helper functions were added to streamline authenticated requests to AI endpoints.
  • UI/UX Improvements: We enhanced the user interface to support seamless interactions with AI components, including real-time predictions, chat interfaces, and analytics dashboards. Bug fixes addressed inconsistencies in AI model outputs and user inputs across sessions.
  • Notification and Email Services: We chose SendGrid for email notifications and FCM (Firebase Cloud Messaging) for push notifications, optimizing these services for reliable communication.
  • Security Enhancements: We reviewed the Supabase structure and implemented necessary security measures, ensuring compliance with data protection standards and preparing documentation for external security audits.

Core Features

Matching User-to-User

Integrate AI to connect users based on shared interests, behaviors, or preferences.

Matching User-to-Event

Use AI to recommend events tailored to user interests and location.

Generate Drink Recipe

Personalise drink recipes with available ingredients and preferences.

Key Results

Improved User Experience

Enhanced UI components and fixed bugs leading to a smoother user experience.

Create Recommendation Algorithm

With unannotated data, use feature extraction and LLMs to generate accurate suggestions.

Similar Search and GPT-4o Model

Analyze and explain matching reasons for among users and user with event.

Use OpenAI Text Model

Leverage OpenAI's text-embedding-large model to convert data into vector representations.

AI playbook for BSFI leaders

Explore high-impact use cases to inspire your AI adoption

Access the playbook now!
Finding an AI solution for your business?

Let us take the hassles away

It’s like having an internal AI experts team helping you to analyse and show where and how to go in the long AI Transformation journey ahead!

Contact Us