As a leading technology company, VNG Cloud manages a diverse product ecosystem, resulting in a massive volume of internal documentation. However, the growing volume of information has become a challenge—employees spend excessive time searching for data, and support teams are overwhelmed with repetitive queries.
To solve this, AI Agents have emerged as the key to streamlining VNG Cloud’s internal processes—from information retrieval and automated responses to improving the employee experience. Let’s explore how we harness the power of AI to transform data into real, actionable intelligence.
Fragmented documents – Complicated operations – Declining productivity
In a corporate environment, data plays a crucial role in daily operations. But without effective organization, it can become a major bottleneck. At VNG Cloud, the internal data—including documents and policies—was scattered and inconsistent. Employees spent significant time searching without gaining real value. This fragmentation didn’t just impact productivity; it also put intense pressure on support teams, who had to repeatedly address the same questions instead of focusing on more strategic tasks. The result: Diminished performance, slower decision-making, and subpar customer experience.
Mr. Nguyen Doan Phu, Head of Products & Partners at VNG Cloud, shared:
“Our Sales and Operations teams wasted a great deal of time looking for product information. Departments must answer the same questions repeatedly. If we had a smart tool, this entire process could be reduced to seconds with just a few keystrokes or clicks.”
This created an urgent need for an automated solution—one that could help employees access the information they need instantly through a unified platform, thereby optimizing workflows and boosting system-wide efficiency.
Turning vision into reality
“We wanted our teams to spend more time on high-impact projects instead of repetitive tasks. So, we made a decision—to leverage AI to streamline information retrieval, boost performance, and enable faster decision-making” - Mr. Phu shared.
While the decision had been made, the path to realizing the idea was filled with challenges.
Overcoming early obstacles: Lessons that paved the way
Initially, VNG Cloud’s AI team aimed to build a powerful virtual assistant—an AI chatbot that could alleviate workload and enhance productivity. But the first results fell short of expectations.
The biggest challenge? Data.
The system had to process outdated, inconsistent, and irrelevant data. Moreover, the input included both structured and unstructured data, complicating how it could be organized and retrieved. This directly impacted the chatbot’s accuracy and performance, requiring significant fine-tuning and optimization.
Even when the chatbot was technically ready, another issue arose: How could non-technical departments adopt it without relying heavily on AI experts? The deployment process was still too complex, hindering scalability and broader adoption. To truly unlock the potential of this technology, the solution needed to be more user-friendly—empowering all teams to tap into AI capabilities, regardless of technical background.
From frustration to functionality: How AI Agent redefined our assistant
Recognizing these limitations, the team went back to the drawing board.
To overcome the early obstacles, VNG Cloud adopted the Retrieval Augmented Generation (RAG) model, allowing AI to deliver precise answers based on enterprise data. At the same time, a system was implemented to automatically clean and standardize inconsistent data, ensuring both accuracy and continuity.
The VNG Cloud team also applied three critical techniques to improve performance and accuracy:
Optimizing retrieval by selecting the right vector database and search algorithm.
Choosing the appropriate Embeddings Model to convert enterprise data into formats AI can understand and process efficiently.
Selecting the right LLM (Large Language Model)—a crucial factor in delivering accurate responses.
Once the chatbot reached a 90% accuracy rate and boosted internal productivity by 30%, VNG Cloud accelerated deployment without depending on engineering teams. With VNG Cloud, in the future, even non-technical departments could build effective chatbots in minutes, using only basic knowledge of AI and data—no coding required.
“AI is not magic, but if built right, it can streamline operations and let people focus on what truly matters” - Mr. Phu emphasized.
VNG Cloud AI Stack – A Comprehensive Ecosystem born from real-world lessons
From the hands-on experience of building an internal AI project, VNG Cloud realized that AI is not just about automation—it’s about optimizing operations and elevating productivity. That’s why we launched the VNG Cloud AI Stack—a comprehensive ecosystem designed to empower every Vietnamese businesses.
AI Agent Studio: A GenAI-powered platform to build enterprise-grade AI chatbots in minutes, with no need for AI expertise.
AI Platform: A robust environment for AI professionals to deploy, fine-tune, and manage AI models effortlessly—without infrastructure overhead.
AI Gateway: An Unified Platform for centralized AI model management that allows businesses to manage hundreds of AI/LLM models on a single platform.
AI Infrastructure: High-performance computing with NVIDIA GPUs, integrated with a Vector Database Platform and VNG Kubernetes Services — optimized for professional AI workloads.
Unlike foreign platforms, VNG Cloud ensures domestic data compliance, cost optimization, and seamless integration. With over 20 years of experience, we provide an end-to-end stack—from data centers and cloud platforms to GPU infrastructure optimized for AI.
Unleash the power of AI for every business
Discover VNG Cloud AI Stack—a comprehensive ecosystem that enables your business to implement AI easily, cost-effectively, and with high efficiency. Whether you're an AI professional or just getting started, we have the right solution for you.
Start today to unlock the power of AI and gain a competitive edge!