OpenAI and Grab AI Skills Programme: Ecommerce Video Implications for Southeast Asia

By VEONIB | 2026-10-14

Quick Answer

OpenAI and Grab launched GO Forward with AI on 2026-09-23, a two-year regional skilling programme that will train 30,000 Grab driver-, delivery- and merchant-partners across Southeast Asia to apply ChatGPT and ChatGPT Work to real business decisions. It begins in Singapore, expands to Thailand, Indonesia and the Philippines later in 2026, then to Malaysia and Vietnam in 2027. The announcement covers AI literacy and agentic task work, not AI video or creative production.

TL;DR

Table of Contents

According to Grab and OpenAI bring practical AI skills to Southeast Asia published by OpenAI on 2026-09-23, the two companies have launched GO Forward with AI, a two-year regional skilling programme intended to help 30,000 Grab driver-, delivery- and merchant-partners apply AI to everyday business decisions. The programme opens in Singapore and expands across Thailand, Indonesia, the Philippines, Malaysia and Vietnam through 2027. For ecommerce operators, the announcement matters less as a product launch than as a signal about where AI literacy is heading: toward merchant-side operations such as sales analysis, promotion planning and storefront building. This article breaks down what was announced, what the underlying survey data shows, how ChatGPT Work fits into agentic merchant workflows, and where the programme connects — or fails to connect — to AI-powered product video production for Shopify, Amazon, TikTok Shop and WooCommerce sellers.

Hero Image Alt Text: Grab and OpenAI GO Forward with AI programme launch for Southeast Asian merchants and driver-partners Caption: The GO Forward with AI launch in Singapore, where OpenAI and Grab committed to training 30,000 partners across Southeast Asia. OG Image Title: OpenAI and Grab Launch GO Forward with AI: What It Means for Ecommerce Sellers Suggested Visual: A split composition showing a Southeast Asian merchant reviewing a dashboard on a laptop beside a delivery driver using ChatGPT on a phone, with a subtle map of Southeast Asia in the background.

What the GO Forward with AI Programme Actually Delivers

GO Forward with AI is a two-year regional skilling programme from OpenAI and Grab that will help 30,000 Grab driver-, delivery- and merchant-partners build practical AI skills for their work and businesses. The deliverables are concrete rather than abstract: half-day, in-person workshops adapted per market, taught from the OpenAI Academy curriculum, plus train-the-trainer sessions so GrabAcademy staff can extend the programme independently over time.

Original Fact

The programme begins in Singapore and expands to Thailand, Indonesia and the Philippines later in 2026, followed by Malaysia and Vietnam in 2027. It forms part of OpenAI for Singapore, the company's broader commitment to working with local partners on AI adoption.

Original Fact

Workshop content is organised around decisions participants already face. A driver or delivery partner might use ChatGPT to explore a business idea, think through what would make it stand out and build a plan. A merchant might use it to look for patterns in sales, plan promotions or prepare for demand changes. In hands-on sessions, participants also try ChatGPT Work on larger tasks — turning sales data into a dashboard, building a simple website or developing an expansion plan — then review the output and make the decisions that follow.

The framing is deliberately human-in-the-loop. OpenAI and Grab describe the programme as building confidence rather than replacing judgement, which matters for merchants whose margins depend on inventory and pricing calls.

VEONIB Insight

The programme's scope is operational, not creative. It teaches merchants to interpret data and plan promotions — the decisions that precede content production — but stops short of teaching asset creation. That is a meaningful boundary for ecommerce teams: a merchant trained to identify a demand spike still needs a separate workflow to produce the product videos that capture it. Businesses should treat this announcement as a signal that AI literacy is becoming a baseline expectation among Southeast Asian sellers, and plan a content production layer to sit on top of that literacy. Adopting the reasoning habits now, while treating video generation as a distinct capability to evaluate separately, is the pragmatic sequencing.

What Grab's Survey Data Reveals About Regional AI Adoption

The programme's design rests on adoption data Grab collected from its own driver-partners in Singapore. Half of those surveyed said they already use AI tools. Among those who do not, 87% said they were open to trying them. ChatGPT was the tool they knew best, and many respondents indicated they wanted to grow an existing business or explore starting one.

Original Fact

OpenAI has worked with Grab since 2024 on tools including the Driver AI Assistant, which is powered by OpenAI models and helps drivers find demand, make the most of their time and answer everyday questions. It has reached close to 500,000 drivers.

Those two numbers sit at different scales. A 50% self-reported usage rate describes intent and habit; a 500,000-driver reach describes deployed infrastructure. Together they suggest the region's constraint is not willingness to use AI but the absence of structured, task-specific guidance.

Suggested visual: a simple two-bar chart comparing the 50% of surveyed Singapore driver-partners already using AI tools against the 87% of non-users open to trying them.

VEONIB Insight

The 87% openness figure is the most commercially important number in the announcement. It describes a large population that has already accepted AI conceptually but has not converted that acceptance into repeatable workflows. For ecommerce tooling vendors, that is a textbook onboarding problem rather than a persuasion problem: the market does not need convincing, it needs a first successful outcome within minutes. The same logic applies to AI video adoption among Shopify and TikTok Shop sellers, where trial rates are high but sustained usage depends on whether the first generated product video is usable without heavy editing.

ChatGPT Work and the Shift Toward Agentic Merchant Operations

ChatGPT Work is the agentic layer the programme relies on for multi-step tasks. Rather than answering a single question, it is described as handling larger jobs — converting sales data into a dashboard, building a simple website, drafting an expansion plan — with the participant reviewing the result and making the final call.

Original Fact

Oliver Jay, Managing Director, International at OpenAI, framed the programme around helping merchants and platform workers gain confidence in using agentic AI capabilities such as ChatGPT Work. Ong Chin Yin, Chief Organisational Capability Officer at Grab, described the goal as helping partners apply AI in ways relevant to their work, businesses and aspirations.

Agentic systems of this kind typically combine an LLM as the reasoning core with tool access, structured inputs and retrieval over the user's own data. The practical implications for merchants are threefold: tasks must be specified precisely, source data must be clean, and outputs must be verified before they reach customers.

VEONIB Insight

Agentic workflows and generative creative workflows are converging, and merchants should understand the seam between them. An agent that reads sales data and drafts a promotion calendar produces structured text; a video pipeline that turns a product URL into a script, storyboard and rendered clip produces media. The two can be chained — agent output can inform which products get video creative first — but they require different tooling, different quality checks and different review cycles. Teams that conflate the two tend to overestimate what a chat interface can deliver for media production. Building an explicit handoff point between planning agents and production tools prevents that confusion.

Rollout Timeline, Markets and Training Model

The rollout is staged by market, with each workshop adapted to local needs and the type of work participants do. The table below summarises the phases described in the source.

Phase Markets Timeline Delivery model
Launch Singapore From 2026-09-23 Half-day, in-person workshops
Expansion wave 1 Thailand, Indonesia, Philippines Later in 2026 Locally adapted OpenAI Academy curriculum
Expansion wave 2 Malaysia, Vietnam 2027 Locally adapted curriculum
Capacity building Regional Ongoing GrabAcademy trainers trained to deliver the programme

Original Fact

OpenAI and Grab state that workshops are based on the OpenAI Academy curriculum and adapted to partners' needs in each country, and that GrabAcademy trainers will be trained so the programme can reach more people over time.

The train-the-trainer component is the part most likely to determine long-term reach. A 30,000-participant target delivered only through OpenAI-led sessions would depend on continued external staffing; transferring delivery capability to GrabAcademy converts the programme into internal infrastructure.

VEONIB Insight

Train-the-trainer models are the difference between a campaign and a capability. For ecommerce agencies serving Southeast Asian merchants, this is a preview of how AI skills will spread: not through vendor-led webinars alone, but through existing operational networks — delivery fleets, merchant communities, platform academies. Agencies that position themselves as the content-production counterpart to these literacy programmes can capture downstream demand. Merchants in the wave-two markets of Malaysia and Vietnam have roughly a year before formal training reaches them, which is a reasonable window to pilot AI video tooling internally and arrive at the workshops with production questions rather than basic ones.

Where Ecommerce Video Fits — and Where the Programme Stops

AI video, image generation and advertising creative production are not mentioned in the source announcement. Not specified in the original source: any curriculum module covering video generation, image generation, avatar tools, voice synthesis or marketing asset production. The stated use cases are business planning, sales analysis, promotion planning, dashboards, simple websites and expansion strategy.

That omission is not a criticism of the programme; it reflects a deliberate decision to teach decision-making before production. But it creates a practical gap for merchants. A Grab merchant-partner who learns to identify a demand pattern still needs to produce the product listing images, short-form ads and product demo videos that convert that demand. In the VEONIB workflow — Product URL → Product Analysis → Script → Storyboard → Image Prompt → Video Prompt → AI Video → Voice → Subtitle → Publishing — the programme covers the analytical front end and stops before the script stage.

The table below is VEONIB's own product assessment of tools a Southeast Asian seller could pair with the programme. It is analysis, not information from the source.

Tool or platform Creative strengths Creative limitations Recommended ecommerce video use
ChatGPT and ChatGPT Work (OpenAI) Structured reasoning, script and storyboard drafting, prompt writing, data-to-dashboard tasks No native video rendering; output requires review Product analysis, video scripts, storyboards, image and video prompts
Sora (OpenAI) Cinematic text-to-video generation Not presented in the source as a merchant self-serve catalogue tool Brand story clips, concept visuals
Gemini and Veo (Google AI) Multimodal understanding combined with video generation Text rendering inside generated video remains unreliable; shot complexity affects consistency Short product ads, YouTube Shorts
Claude (Anthropic) Long-context reasoning over product documentation No video generation capability Product research, copy review, script editing
Runway (Runway) Broad generation and editing toolkit with strong motion control Requires operator skill; per-second costs accumulate at volume Product ads, lifestyle sequences
Pika (Pika) Fast, stylised short-form generation Less suited to precise product demonstration TikTok-style hooks, social cutdowns
HeyGen (HeyGen) Avatar-led presenter video and localisation Avatar realism varies; needs brand-safety review UGC-style ads, how-to videos, multilingual variants
Hailuo (MiniMax) Fast short-clip generation Limited long-form narrative control Social teasers, motion loops
Seedance (ByteDance) Generation aligned with short-form social formats Availability tied to ecosystem access TikTok Shop creative

When evaluating any of these for ecommerce, the criteria that matter most are product consistency, character consistency, text rendering quality, camera movement capability and prompt controllability. Production speed and cost efficiency determine scalability; motion quality and visual fidelity determine whether the result can carry a paid campaign.

VEONIB Insight

The realistic near-term role for programme graduates is upstream, not downstream. Merchants who can articulate a product angle, a target audience and a promotion window produce better video prompts than merchants who cannot — and prompt quality is the single largest determinant of usable output in AI video generation. Businesses should adopt agentic planning tools now and treat video generation as a separate, staged investment: pilot one product category, measure cost per usable clip, then scale. Waiting is preferable only where a merchant has fewer than roughly twenty SKUs and no repeat ad spend, since manual production remains competitive at that volume.

Risks, Gaps and What the Announcement Does Not Specify

Several practical details are absent from the source. Not specified in the original source: the cost of the programme to participants, whether completion carries any certification, how learning outcomes will be measured, which languages workshop materials will be delivered in, and how the 30,000-participant target compares with Grab's total partner base in the region.

The measurement gap is the most consequential. Skilling programmes routinely report enrolment and reach, which are output metrics, while the business outcomes that justify them — revenue growth for merchant-partners, time saved on administrative tasks, new businesses launched — are harder to track and rarely disclosed.

VEONIB Insight

There is also a concentration risk worth naming. A single superapp partnering with a single model provider creates a training pathway that is tightly coupled to one ecosystem. Merchants who learn ChatGPT-centric workflows may find their skills less portable if they later adopt Google Gemini, Microsoft Copilot or Meta AI tooling. The practical mitigation is to teach transferable skills — structured prompting, data hygiene, output verification — alongside specific interfaces. Businesses evaluating their own training should weight conceptual portability over menu-level familiarity.

Outlook: From AI Literacy to AI Content Production

The programme's two-year horizon maps onto a broader regional shift. Southeast Asia has high AI tool awareness and comparatively low workflow integration; programmes like GO Forward with AI target exactly that gap. As literacy rises, the constraint on merchant growth moves from "not knowing what AI can do" to "not having the production capacity to act on it."

The natural next step for merchants is the creative layer. Product pages, marketplace listings and short-form social ads all require visual assets, and the volume required per campaign keeps rising across TikTok Shop, Amazon and Shopify channels. Tools that reduce the distance between a product URL and a publishable video — rather than adding another manual step — are the ones most likely to see sustained adoption in this region.

VEONIB Insight

The strategic takeaway is that AI literacy programmes create demand for AI production tools rather than satisfying it. Every merchant trained to plan a promotion faster will eventually need to produce the creative for it faster too. For founders and agency operators, the commercially interesting window is 2026 to 2028, when literacy spreads ahead of production capability. Positioning a production workflow now — measured in cost per finished video, not in features — is likely to matter more than waiting for the training market to mature.

Recommendations

Shopify Merchants

Start with the analytical steps the programme teaches: use AI to review sales patterns and select the two or three products with the clearest margin and demand case. Then produce video creative for those SKUs only. Measure cost per usable asset before expanding to the full catalogue.

Amazon Sellers

Focus on the listing-video slot and A+ content requirements, where product consistency matters more than cinematic quality. Test one AI video pipeline against your existing studio process for a single ASIN, then compare conversion rate and cost per unit.

AI Developers

Build for the handoff between agentic planning and media production. Structured inputs — product URLs, catalogue feeds, structured sales exports — reduce prompt ambiguity and improve output consistency more than model upgrades alone.

SaaS Founders

The programme's 30,000-participant target defines an addressable user base that will be trained on AI workflows but not on AI production. Position products as the production counterpart to literacy training, and price against the cost of a freelancer-produced video, not against software subscriptions.

Content Marketers

Treat workshops and community programmes as distribution channels. Localised, market-specific case studies — Singapore, Indonesia, Vietnam — will outperform English-language generic material in these markets.

Video Creators

Shift effort from first-draft generation, which is increasingly automated, toward art direction, product accuracy review and brand consistency. These remain the tasks where human judgement produces measurable quality gains.

FAQ

What is the GO Forward with AI programme? It is a two-year regional skilling programme from OpenAI and Grab that will help 30,000 Grab driver-, delivery- and merchant-partners across Southeast Asia build practical AI skills for their work and businesses.

Which countries does the programme cover? It launches in Singapore, expands to Thailand, Indonesia and the Philippines later in 2026, and reaches Malaysia and Vietnam in 2027.

Which AI tools does the programme teach? Participants work with ChatGPT for everyday tasks and ChatGPT Work for larger multi-step jobs, based on the OpenAI Academy curriculum.

Does the programme include AI video or image generation training? Not specified in the original source. The stated use cases are business planning, sales analysis, promotion planning, dashboards, simple websites and expansion plans.

Is the programme free for Grab partners? Not specified in the original source. Cost, certification and outcome measurement are not disclosed in the announcement.

How does this affect ecommerce sellers outside Southeast Asia? The main implication is directional rather than operational. It signals that AI literacy is becoming a baseline expectation among marketplace sellers, which raises the baseline for creative production tooling globally.

References

Sources

Try VEONIB

VEONIB converts a product URL into product analysis, video scripts, storyboards, image prompts, video prompts and finished AI marketing videos automatically. Merchants who have completed AI literacy training can use it to move directly from promotion planning to publishable creative.

Credibility Assessment

This article separates three layers of information. Facts drawn directly from the source include the programme name, the 30,000-participant target, the two-year timeline, the market rollout sequence, the survey figures (50% usage, 87% openness), the Driver AI Assistant's reach of close to 500,000 drivers, the use of the OpenAI Academy curriculum, the train-the-trainer model and the attributed quotes.

Analysis produced by VEONIB includes the assessment of the literacy-to-production gap, the evaluation of agentic planning versus generative creative workflows, the tool comparison table, the adoption sequencing recommendations and the 2026–2028 outlook. These are judgements based on industry patterns, not statements from OpenAI or Grab.

Uncertain or unavailable information includes programme cost, certification, language coverage, outcome measurement methodology and any curriculum content relating to video, image or creative asset production. These points are explicitly marked as not specified in the original source rather than inferred.