30,000 Drivers, Merchants, and One ChatGPT Curriculum: Inside Grab and OpenAI's 'GO Forward with AI'

On September 23, Grab and OpenAI announced “GO Forward with AI,” a regional partnership to equip 30,000 driver-, delivery-, and merchant-partners across Southeast Asia with practical AI skills over the next two years. The programme launched first in Singapore under GrabAcademy, with expansions to Thailand, Indonesia, and the Philippines planned for later this year. The name is a nod to the two companies behind it — “GO” stands for Grab and OpenAI — and, according to the announcement, for the momentum behind their work together.

It is OpenAI’s first regional AI-skilling programme in Southeast Asia, and it marks a notable shift in how frontier AI labs approach emerging markets: not by shipping a product and waiting for adoption, but by physically showing up — half-day, in-person masterclasses, market by market — to teach gig-economy workers how to put agentic AI to work.

A partnership that did not start last week

The collaboration between the two companies dates back to May 2024, when Grab began working with OpenAI on applied-AI initiatives. The most visible product of that relationship is Driver AI Assistant, Grab’s in-app assistant for drivers that provides personalised guidance to find demand, optimise earnings, and resolve day-to-day questions. Powered by OpenAI’s models, the assistant has reached close to 500,000 drivers — already one of the largest deployments of AI assistance to platform workers anywhere in the world.

“GO Forward with AI” extends that track record from in-app tooling to human capability. Rather than building another assistant, the programme aims to build users: partners who can delegate multi-step work to ChatGPT on their own.

Teaching agents, not chatbots

The curriculum centres on ChatGPT Work and its agentic capabilities. Instead of treating AI as a one-off question-answering machine, participants learn to delegate multi-step tasks: creating and executing a business expansion plan, turning sales and forecast data into an interactive tracking dashboard, building a simple website to showcase products, or creating repeatable workflows for everyday operations like inventory management — “while remaining in control of the outcome,” as the announcement emphasises.

The training is deliberately split by audience. For driver- and delivery-partners in Singapore, the focus is entrepreneurship: using ChatGPT to shape and validate business ideas, sharpen value propositions, explore pricing approaches, and produce simple marketing materials and action plans. For merchant-partners, the emphasis is on smarter business decisions and daily operations — analysing point-of-sale data, identifying trends, shaping revenue and promotional strategies, and building AI-powered workflows for menu engineering, demand forecasting, inventory planning, and waste reduction. These capabilities complement the AI tools already on Grab’s merchant platform, including the Merchant AI Assistant.

The programme is built on OpenAI Academy’s curriculum, tailored for Grab’s partner communities, and delivered through a train-the-trainer model: OpenAI and Grab teams will visit participating markets to run the masterclasses while equipping GrabAcademy’s regional trainer network to scale the programme further over time. Every participant also receives three months of complimentary ChatGPT Plus access to keep practising after the workshops end.

The adoption gap the programme is attacking

The numbers Grab disclosed explain why the companies felt a training programme was needed at all. In Singapore, only 50% of driver- and delivery-partners currently use AI tools. Among those who have yet to adopt them, 87% say they are open to doing so. The gap is even wider on the merchant side: only 33% of merchant-partners currently use AI in their work, and only 25% say their employees can apply AI effectively.

In other words, willingness is not the bottleneck — practical know-how is. “Our partners are intrigued by AI, and many are looking for practical guidance on how to apply it in ways that are relevant to their work, businesses and aspirations,” said Ong Chin Yin, Chief Organisational Capability Officer at Grab. “This programme is about building that confidence — whether it’s helping a merchant make smarter business decisions or giving a driver-partner the tools to explore a new business idea and create new opportunities for growth.”

Oliver Jay, Managing Director, International at OpenAI, framed the programme as part of the company’s accessibility commitments: “ChatGPT has seen strong adoption across Southeast Asia, reflecting the region’s openness to new technology, but many are still only scratching the surface of what it can help them achieve. Launching the ‘GO Forward with AI’ programme first in Singapore is part of our commitment to make AI accessible to all.”

Early results, in the partners’ own words

The programme was developed and shaped by Grab partners in Indonesia, Singapore, and Thailand who aspire to grow or start businesses but lacked the practical digital skills to turn those ambitions into reality. The announcement includes testimonies that hint at why the hands-on format matters.

Manfred Ho, founder and CEO of Genista Lane, described connecting ChatGPT Work directly to his payroll files and completing “in 20 minutes what normally takes two hours.” A driver-partner who attended the Singapore workshops described moving “from idea to execution” — exploring new ways to support a family hawker business while building projects that could generate alternative streams of income.

These are modest, concrete outcomes — which is precisely the point. The programme’s theory of change is that AI adoption among platform workers will not come from flashy demos but from payroll done faster, menus engineered better, and side businesses launched with an AI collaborator rather than a consultant.

Why this matters beyond Southeast Asia

The Grab-OpenAI programme is worth watching for three reasons. First, it is a template for how AI labs can deepen their footprint in growth markets: OpenAI is competing for the next billion users largely on distribution and localisation, and embedding its curriculum inside a super-app’s training academy is a form of distribution that money alone cannot easily buy. Second, it is an early large-scale test of whether “agentic” AI — multi-step task delegation rather than single-turn chat — can be taught to non-technical, non-office workers, and whether that translates into measurable income or efficiency gains. And third, it feeds a growing body of evidence about AI and gig work at a moment when regulators and unions are scrutinising exactly this intersection.

If the train-the-trainer model scales as intended, 30,000 partners is only the beginning. If it stalls, it will be a data point that the skills gap — not the technology — is the binding constraint on AI adoption in the real economy. Either way, the next two years across Singapore, Thailand, Indonesia, and the Philippines will offer one of the clearest natural experiments yet on teaching agentic AI to the workforce that keeps Southeast Asia moving.