• Grab’s AI Centre is building voice tools for visually-impaired users.
• It’s also training local talent to develop AI for Southeast Asia.

Grab has opened its first Artificial Intelligence Centre of Excellence (AI COE) at its Singapore headquarters, which is backed by Digital Industry Singapore (DISG). The centre will develop AI solutions to improve access to digital services, boost productivity, and support public projects across Southeast Asia.

Grab will use the new facility to train local talent, build tech tools for everyday use, and address regional social and business concerns. By 2025, the company expects to create at least 50 roles in engineering, product, data science, and analytics. These jobs are aimed at helping Singaporeans work on real AI problems with regional impact.

One major focus is making technology easier to use, especially for people who have trouble with standard apps or services. Grab is working with the Singapore Association of the Visually Handicapped (SAVH) to test a new voice assistant that helps visually-impaired users book rides using spoken commands. SAVH members helped test the tool and gave feedback to make sure it fits their needs.

Lyn Loh, who leads accessibility services at SAVH, said they were glad to support this effort. She said the voice assistant could help more people travel on their own and use digital services more easily.

Grab built the voice feature using models based on OpenAI’s technology, but with a local twist. The team trained the system on 80,000 voice samples that reflect Singaporean speech and building names. That helped the model improve its ability to recognise local accents and landmarks—from 46% to 89%. Starting in June, users in Singapore can donate voice samples through the app to further improve accuracy.

For now, the voice assistant is being tested in Singapore. Grab plans to expand its use in the future and is also exploring similar tools to support elderly users and others less familiar with smartphones.

The AI COE will also work on boosting productivity. That includes assisting drivers, delivery workers, and small businesses in using AI to make better decisions and improve how they work.

Grab is building its own AI model, trained on its own data, to better understand user behaviour and how partners use the app. This custom model is expected to improve the way Grab recommends rides, routes, and services to users and partners. It can help provide suggestions that are more relevant to each individual.

The company is also creating tools that let its employees test AI ideas more quickly. A new internal kit includes code and resources that make it easier to set up safe environments for AI testing. What used to take more than a week can now be done in half a day.

For Grab’s partners, the AI COE is rolling out tools like the Driver AI Companion and the Merchant AI Assistant. The driver tool offers tips on where to go for more ride requests, how to plan routes more efficiently, and what rewards are available based on driving patterns. The merchant tool assists businesses in understanding trends, getting suggestions on improving sales, and managing daily tasks more easily.

Another area of focus is public infrastructure and smart city tools. Grab collects a lot of data from its driver network, which it plans to use to support local government work. Its existing devices, such as KartaDongle and KartaDashcam, are being updated with new features to detect road hazards, monitor weather, and track traffic in real time.

One example is its collaboration with Singapore’s national water authority, PUB. By merging Grab’s data with PUB’s flood monitoring systems, the two teams can deliver real-time flood notifications to drivers and assist officials respond more quickly to storm-related situations. This reduces traffic interruptions and improves emergency response times.

Grab describes the AI COE as a way to use technology to improve daily life across Southeast Asia. The centre brings together data, tools, and people to build useful solutions for drivers, merchants, users, and cities. By focusing on real problems like accessibility, small business support, and urban planning, the company hopes to make AI more practical and inclusive across the region.