Grab Building 'Personalized Concierge' AI
Grab's COO revealed the company is building a suite of proprietary AI agents for personalization, semantic search, and predictive intent. The super app uses over 1,000 internal models to transform its service into a "personalized concierge" for users, with a showcase planned for its upcoming Product Day.
Grab's push into AI is guided by a philosophy of "AI-First with Heart," aiming to use technology to solve real-world problems for its users, partners, and the community. This strategy was emphasized by Group CEO Anthony Tan at the inaugural GrabX product showcase, where the company's latest AI-powered innovations were unveiled. The company's commitment to AI includes a nine-week, company-wide "G/AI Sprint" to immerse all employees in generative AI. This initiative is not just about customer-facing features; Grab is also deploying "agentic AI" to empower its merchant and driver-partners. Developed in collaboration with OpenAI and Anthropic, these tools act as intelligent assistants. The AI Merchant Assistant provides 24/7 business advice on everything from menu optimization to advertising, while the AI Driver Companion helps drivers find high-demand areas and report traffic conditions with their voice. To power these features, Grab has built its own user foundation model on a transformer architecture. This allows the company to unify different types of user data to create a more holistic understanding of its customers. Instead of building large language models from scratch, Grab partners with companies like OpenAI and Anthropic and fine-tunes their models with its own extensive, hyperlocal datasets. This approach allows them to better capture the nuances of Southeast Asian languages and user behaviors. The company's focus on AI extends to an internal "AI Gateway" that gives employees access to a range of external AI providers like OpenAI, AWS Bedrock, and Google VertexAI, as well as in-house models. This centralized platform is designed to accelerate innovation while maintaining security and managing costs. Internally, AI tools like "Jarvis" have already saved the company over 30,000 hours of manual work by allowing employees to retrieve data using plain English queries. Looking forward, Grab plans to expand its use of AI to further segment and cater to diverse user needs, with features like "Grab for Family | Teens" and "Shared Saver". The company is also exploring AI-driven loyalty tools and using AI to enhance its financial services by underwriting loans more precisely than traditional banks. This aggressive push into AI is a core part of Grab's strategy to triple its EBITDA to $1.5 billion by 2028.