Enables centralized AI model training and deployment with Vertex AI
Accelerates multilingual AI model development speed by 70% with Gemini
Helps boost user satisfaction rate for customer service by 5-10%
Enhances customer service efficiency by 30% with Vertex AI Search
With Google AI, AIHelp provides AI customer service agents with higher response accuracy rates across languages.
With Google AI, AIHelp provides AI customer service agents with higher response accuracy rates across languages.
Building quality multilingual AI customer service agents
We compared several offerings and found Google Cloud the best solution to build multilingual AI agents with minimal hallucinations. Besides, it provides a centralized AI platform that enables streamlined MLOps workflows, significantly reducing costs.
Steven Sun
Co-founder, AIHelp
AI agents are becoming a prerequisite for customer service, because their 24/7 availability enables customers to get answers quickly anytime. The use of AI customer service agents has grown by more than 20X in 2025, with 94% of customers opting for AI agents when available.
AIHelp has been providing AI-powered customer service to enterprise customers since 2014. Tailored for the gaming industry, its customized AI solutions have evolved from FAQ chatbots to AI agents capable of multi-round reasoning and intelligent interaction. That's thanks to the adoption of robotic process automation (RPA), natural language processing (NLP), machine learning (ML) and large language models (LLMs).
"Gaming is a complex use scenario for AI customer service agents, because users raise questions frequently, and each game has its own unique rules and gameplay," explains Steven Sun, co-founder of AIHelp. "With our rich experience in the sector and the ever-evolving AI technology, we're dedicated to helping game publishers and developers continuously improve the efficiency and quality of their customer service."
To serve its enterprise customers in 32 countries, AIHelp's AI customer service solutions are available in more than 40 languages. As its business expanded, AIHelp noticed that its AI agents had limited comprehension capability in certain languages, and hallucinations frequently appeared in more complex scenarios, leading to low response accuracy rate. Aiming to make its AI agents useful in more languages and use cases, AIHelp looked for AI model alternatives and chose Google Cloud for its multilingual NLP capabilities and developer-friendly AI tools.
"We compared several offerings and found Google Cloud the best solution to build multilingual AI agents with minimal hallucinations," notes Sun. "Besides, it provides a centralized AI platform that enables streamlined MLOps workflows, significantly reducing costs."
Enhancing customer service quality while speeding up AI development with Google AI
Following the Google Cloud team's architecture advice, AIHelp employs Vertex AI as its centralized MLOps platform. This way, AIHelp is able to merge the AI tools of Google Cloud and its self-developed agentic AI capabilities into a streamlined workflow that automates model training, testing, and deployment, preventing operational errors.
With this new MLOps platform, AIHelp has improved its AI agent's response accuracy rate. The team uses Gemini as its core LLM to support multi-round conversations, complex intent reasoning, and context memory. As Gemini's wide context window can process full user information to match elements in AIHelp's knowledge base, it generates responses that contain more detailed information, resulting in higher accuracy.
Gemini is a powerful LLM with excellent comprehension and reasoning capabilities. What's even better is that we can easily take advantage of other AI tools on Vertex AI to optimize Gemini's responses in multiple languages and for complex use cases.
Steven Sun
Co-founder, AIHelp
Meanwhile, Translation AI ensures that prompts and responses are translated correctly and naturally, while the RAG engine of Vertex AI enables vector search that can effectively retrieve the most relevant content from the knowledge base, minimizing hallucination risks. To further enhance response quality, AIHelp also uses the summarization, paraphrasing, and translation evaluation features of Vertex AI to automatically rewrite and evaluate responses generated by Gemini.
"Google AI has helped our AI customer service agents make a huge leap in performance," says Sun. "Gemini is a powerful LLM with excellent comprehension and reasoning capabilities. What's even better is that we can easily take advantage of other AI tools on Vertex AI to optimize Gemini's responses in multiple languages and for complex use cases."
On top of that, merging prompt comprehension, translation, and quality assurance into one single workflow has boosted AIHelp's development efficiency. Before the adoption of Google Cloud, AIHelp's DevOps team needed several days to build a multilingual question-and-answer AI model, because it had to create knowledge bases and prompt scripts in multiple languages separately. Now, it only requires 70% less time.
Expanding AI agent's decision-making capacity for better customer service
With Google Cloud, we're confident that we can continue upgrading our AI-powered customer service and help realize instant, thoughtful customer care around the world.
Steven Sun
Co-founder, AIHelp
AIHelp's updated AI-powered customer service has received positive feedback from its enterprise customers. Since AIHelp's AI agents now require less human review and can resolve more users' questions in the first round of conversation, its customers have seen a 5-10% increase in their user satisfaction rate. Through Vertex AI Search, AIHelp also empowers customer service personnel to find answers in documentation faster, leading to 30% higher case handling efficiency.
Next, AIHelp will add more automated customer service with Gemini's multimodal capabilities to handle deposit, refund, report reviews, and bug identification faster. It will also use the RAG and matching engine of Vertex AI to build dynamic knowledge bases, and develop AI agents with higher decision-making capacity, as well as AI assistants for human customer service agents.
"We're stepping into a new era when AI agents are becoming more reliable and bringing substantial improvements to customer service," notes Sun. "With Google Cloud, we're confident that we can continue upgrading our AI-powered customer service and help realize instant, thoughtful customer care around the world."
AIHelp provides localization and AI-powered customer service to game developers and publishers. It currently serves thousands of enterprise customers who have more than 800 million monthly active users.
Industry: Gaming
Location: Singapore
Products: Google Cloud, Gemini, Translation AI, Vertex AI, Vertex AI Search
