VinBigData fuels Vietnam’s innovation with AI and MongoDB

INTRODUCTION

VinBigData builds a range of high-tech artificial intelligence (AI) products for customers across healthcare, financial services, logistics, and the automotive industry in Vietnam. A subsidiary of Vingroup, one of Vietnam’s largest conglomerates, VinBigData’s solutions include, VinBase - a generative AI platform, and Vizone - an ecosystem of intelligent image analysis solutions.

In 2022, VinBigData launched VinBase, a comprehensive multi-cognitive AI platform to help companies across diverse industries use AI to optimize production, improve business activities, and build a better user experience. Built on MongoDB Atlas, VinBase includes speech and natural language processing, speech recognition, and speech synthesis products like the VinBase Chatbot, the VinBase Callbot, the ViVi Virtual Assistant, and all the VinBase APIs.

These VinBase services have helped diverse customers across Vietnam, like Asia Commercial Joint Stock Bank (ACB), build AI-powered solutions, including a customer service support chatbot that interacts with millions of the bank’s customers quickly, accurately, and naturally. Automaker VinFast is currently integrating the ViVi Virtual Assistant into the company’s smart electric car lines, helping drivers perform many “hands-free” tasks on the go, including smart car control and navigation in natural language.

“At VinBigData, we leverage our deep expertise in AI and data science to create innovative solutions that enhance the user experience,” said Viet Ha Nguyen, Head of AI Platform Development, VinBigData. “With MongoDB, we have built a suite of AI-driven products that showcase our commitment to building customer-centric solutions on the cloud.”

THE CHALLENGE

Fragmented data across multiple business units

VinBigData needed to support its customers, including parent company Vingroup, in centralizing and integrating often fragmented data and AI solutions to drive efficiencies and stimulate innovation.

The lack of a centralized data management system made it challenging for VinBigData’s customers to maintain data accuracy, while monitoring and controlling operations effectively.

“Fragmented data across multiple business units stifled our customers’ ability to innovate, expand, and make data-driven decisions. We built a centralized data management solution on MongoDB to provide our customers with a comprehensive, unified, and real-time view of their operations,” said Nguyen.

THE SOLUTION

VinBigData first built its applications on MongoDB for its database features and in 2022 the company migrated to MongoDB Atlas on Amazon Web Services (AWS), as VinBigData’s application services’ architecture was also built on AWS. With MongoDB Atlas, VinBigData’s customers can easily integrate and analyze petabytes of historical and real-time data from their operations. This helps companies like VinFast gain insights to improve decision-making and accelerate its global electric vehicle ambitions with improved product and service quality, while enhancing user experiences.

MongoDB Atlas also provides VinBigData with a scalable, stable, cloud-native, fully-managed and secure database to easily deploy and expand its AI applications, including chatbots and speech recognition solutions that fulfil customers’ needs.

MongoDB's flexible schema and data model allows VinBigData to work with different data types intuitively and rapidly change its application service solution in step with its ever-evolving products. With VinBigData’s foundation layer (modern, multi-cloud database platform, MongoDB Atlas) ready for a new deployment and able to process unfamiliar data at high performance, VinBigData can deliver a new AI product in as little as two weeks.

THE RESULTS

Database operations cost drop 30%, API latency remains below 100ms

With MongoDB, VinBigData reduced infrastructure operations and management costs by 30%, without requiring a dedicated infrastructure team. Intuitive and easy to use, MongoDB Atlas, with supporting features like Performance Advisor, and comprehensive documentation, helped engineers and developers to proactively detect and mitigate potential issues like performance degradation due to abnormal peaks in record creation, keeping API latency consistently below 100ms. Low latency in APIs is important because it helps ensure a smooth and fast user experience, which is critical for customers offering online gaming, streaming, or financial trading solutions. VinBigData has started to experiment with building generative AI products with MongoDB Atlas Vector Search. This service allows VinBigData to search and retrieve relevant data based on meaning and context, rather than just keywords, helping the company build intelligent search engines, recommendation systems, and question-answering applications powered by generative AI models at low latency.

Find out more about how MongoDB Atlas’ integrated suite of cloud database and data services can accelerate and simplify how you build with data.