Grab CEO Anthony Tan is restructuring the Singapore-headquartered super-app around what he calls a cyborganisation — a model in which every employee works alongside AI tools to make operational decisions at scale. The push comes as Grab reported 2024 revenue of US$2.35 billion, up 30% year-on-year, and achieved its first full-year adjusted EBITDA profitability of US$150 million, ahead of an earlier 2026 target. More than 90% of Grab’s engineers now use AI coding tools, and the company has run nine weeks of company-wide generative AI experiments since 2024.
The more consequential story is not the internal branding. It is that Grab is deploying AI across credit scoring, routing, and pricing in markets where AI-specific regulation does not yet exist — making it a live test case for how data-rich super-apps behave when regulators are still catching up.
When Anthony Tan halted normal business operations for nine weeks in 2024 and directed every Grab employee — from senior executives to customer service agents — to run generative AI experiments, he was not running an innovation exercise. He was betting that the company’s next competitive advantage would be invisible to users: wired into the back-end systems that price a ride in Da Nang, approve a micro-loan in Kuala Lumpur, or route a delivery courier through central Jakarta.
Tan, named Businessman of the Year at the 41st Singapore Business Awards, frames the destination as a “cyborganisation” — a company in which every employee functions as a human-AI hybrid. The language is deliberately provocative, but the operational logic behind it is straightforward. Grab has accumulated 20 billion transactions across more than 900 cities in eight Southeast Asian markets. That data volume is the raw material; AI is the refinery.
What makes the strategy genuinely significant — and genuinely risky — is the regulatory environment in which it is being executed. Singapore and Malaysia govern Grab’s data practices under Personal Data Protection Acts that are principles-based rather than prescriptive. There is no Southeast Asian equivalent of the EU’s AI Act to define what algorithmic credit scoring or AI-driven worker management must look like. Grab is building the model before the rules arrive.
RELATED
The details: from MyTeksi to machine intelligence
Grab’s origins were deliberately modest. Founded as MyTeksi in Malaysia, the platform’s first product was ride-tracking — a safety feature that allowed passengers to share their journey with a trusted contact. Tan credits that founding constraint, solving a real safety problem for women travelling alone at night, with instilling the data discipline that now underpins the AI push. The company claims 99.9% of all rides across its network are verified as safe, a figure it says is KPMG-audited.
The financial foundation for the AI build-out is now solid. Financial results confirmed in February 2025 showed 2024 deliveries revenue of US$1.36 billion and mobility revenue of US$793 million. Profitability arrived ahead of schedule, driven by cost optimisation and early automation gains. The question investors are now asking is whether AI can expand margins rather than merely protect them.
Tan’s AI copilot ambition extends beyond engineering teams. In a 2025 earnings call, he stated that Grab aims to embed AI copilots into “every key function” from customer service to driver onboarding. The fintech dimension is particularly significant: one in three active Grab drivers currently borrows from the platform, using what Tan calls a hustle score — a proprietary creditworthiness model built on driving behaviour and service ratings rather than conventional credit history. That is the kind of AI application that sits in a regulatory grey zone across the region.
Grab’s acquisition of Foodpanda Taiwan adds 21 new cities to its network, extending the cyborganisation model beyond Southeast Asia for the first time. Taiwan’s data protection framework differs materially from Singapore’s, adding a new compliance layer to what Tan describes as a push to scale “way beyond the borders of Southeast Asia.”
Singapore’s Personal Data Protection Commission requires organisations to obtain consent for collecting, using, or disclosing personal data, limits cross-border transfers to jurisdictions with comparable protection, and mandates breach notification in specified cases. Malaysia’s Personal Data Protection Act 2010 adds registration requirements and data retention standards for commercial platforms including ride-hailing and delivery operators. Unlike the EU’s AI Act, both frameworks rely on high-level principles rather than AI-specific legislation — creating flexibility for Grab but few prescriptive guardrails on high-risk applications like algorithmic credit scoring.The intelligence layer nobody else has
Tan’s most revealing claim is not about AI in the abstract. It is about geography. Grab has mapped more than 900 cities at what he calls “hyper-local” resolution — down to individually numbered beach benches in Da Nang, each logged as a point of interest in the platform’s dispatch system. That granularity, built over a decade of transaction data, is the competitive moat that generic AI models cannot replicate from outside the region.
Rolf Degenkolb, Partner at Bain & Company Southeast Asia, argued in a 2025 Bain commentary that super-apps like Grab and GoTo must use AI to personalise services and optimise driver and courier networks, or risk losing market share to more focused, AI-driven vertical players. The threat is real: Grab holds approximately 75% of Singapore’s ride-hailing market by trip volume as of 2023, but that dominance is a product of network density, not technological lock-in. A well-capitalised vertical competitor with better AI could erode it faster than a decade of super-app expansion built it.
The forward signal to watch is Grab’s full-year 2025 earnings guidance, expected in early 2026, where management has indicated it will quantify cost savings and revenue uplift from AI copilots. Material productivity gains would validate the cyborganisation thesis. Muted results would raise harder questions about the return on nine weeks of halted normal operations and the ongoing investment in multiple R&D centres across Singapore, Vietnam, India, China, and the United States.
Beyond the headline
The bigger picture
Grab’s cyborganisation vision illustrates how AI is shifting from a discrete product feature to the operating system of entire companies, particularly in data-rich platforms with high transaction frequency. The deeper transition underway in Southeast Asia is one where competitive advantage comes less from owning the super-app front end and more from how deeply AI is wired into routing, risk, pricing and workforce management behind the scenes. The company that wins that back-end race may be invisible to consumers but decisive for investors.
The reach
For Western firms, Grab’s experiment demonstrates how AI-saturated service platforms can rewire labour, banking and logistics in emerging markets without the heavy, prescriptive AI rules taking shape in Europe. Western investors and corporate partners in mobility, food delivery and fintech gain an early look at how AI-driven super-apps may compete with — or complement — Western incumbents as they expand into Southeast Asia or export similar platform models elsewhere. The Taiwan expansion, in particular, opens a market with stronger institutional ties to Western technology companies.
Our take
Grab’s push to make every employee effectively a cyborg with AI support is less a gimmick than a logical extension of its data-heavy, thin-margin business. The real risk is not over-automation but under-governance: without AI-specific rules on fairness, transparency and worker impact, the region’s flagship super-app could entrench opaque algorithmic management — particularly in credit scoring and driver performance assessment — long before regulators and civil society are equipped to challenge it.
What this means for investors, travellers, and platform workers
With Grab deploying AI across credit, pricing and workforce management in markets where AI-specific regulation is still absent, the stakes are concrete for several distinct groups.
- Investors tracking Southeast Asian tech: Monitor Grab’s full-year 2025 earnings release, expected in early 2026, for the first quantified disclosure of AI-driven productivity gains. If the cyborganisation is working, management will name the numbers. If it is not, the nine-week operational halt in 2024 will look more costly in retrospect. Grab trades on NASDAQ under the ticker GRAB.
- Western firms considering Southeast Asian platform partnerships: Grab’s R&D centres in Singapore, Vietnam, India, China and the United States make it a potential technology partner as well as a distribution channel. The Taiwan expansion via Foodpanda creates a new entry point for firms already operating in that market.
- Travellers using Grab across the region: The platform’s hyper-local mapping — including fixed pickup points at specific beach benches in Da Nang — makes it operationally more precise than generic mapping apps in many Southeast Asian cities. For practical guidance on using Grab at major transport hubs, the fixed-price Grab pickup system at Manila Airport illustrates how the app’s pricing transparency compares to unmetered alternatives.
- Regulators and policy observers: Singapore’s Personal Data Protection Commission and Malaysia’s Personal Data Protection Commissioner are the primary oversight bodies for Grab’s data practices. Neither jurisdiction currently has AI-specific legislation governing algorithmic credit scoring or automated worker management. The EU’s AI Act, which does impose such requirements, provides the clearest comparative benchmark for what prescriptive oversight could look like.
- Platform workers and driver-partners: Grab’s hustle score model determines loan eligibility for one in three active drivers. Workers who dispute algorithmic assessments currently have limited formal recourse under existing PDPA frameworks, which were designed for data consent rather than algorithmic accountability.
