Cryptocurrency ecosystems currently present a fascinating study in shifting momentum and changing investor psychology. At the time of writing this analysis, the flagship digital token trades at US$63,480.75. Over the last 24 hours, the financial environment experienced microscopic upward movement.

The primary cryptocurrency advanced 0.01 per cent to reach US$63,730. Concurrently, daily transaction volume increased by 1.27 per cent to reach US$22,340,000,000. These marginal shifts mask deeper structural fatigue gripping the broader speculative landscape. Buying momentum weakened significantly throughout August.

I closely monitor the 200-week simple moving average. This specific metric serves as a critical indicator of cyclical bottoms for long-term holders. The benchmark frequently marks accumulation zones during extended bear phases. Prices holding above this threshold typically signal bullish sentiment across the board. A decisive break below this line unleashes severe downward pressure.

The current technical setup places this vital support level under immense strain. Historically, the metric provided strong buying pressure around July. That strength helped prices stage a modest rebound during the summer months. I made a specific forecast six weeks ago that perfectly captured this dynamic. They predicted the benchmark would only provide short-term support in July and trigger a minor bounce.

Those experts cautioned about the distinct risk of a medium-term breakdown. Their warning proved entirely accurate. The digital coin found support last month. August brought no clear buying catalyst to sustain the previous momentum. The effectiveness of that foundational support markedly waned as weeks progressed.

The post Pokemon cards gained 22.8% while Bitcoin lost 20.7% and that gap should worry every investor appeared first on e27.

For all the excitement around AI-assisted coding, a stubborn gap remains: building a demo has become easier, but turning that demo into a reliable product is still hard.

That is the problem GENCOW, a South Korea-based AI service development platform, is trying to solve. Founded by Kevin Choi, a former Samsung Electronics executive with more than 15 years of experience building global software products, the company sits in a fast-growing category of tools that promise to help founders and developers move from idea to working application with less backend engineering.

Also Read: No fans, no fridges, just paint: ZERC’s founder on cracking SEA’s cooling crisis

Choi’s view is blunt: AI will not make developers obsolete. Instead, he believes it will create many more of them.

“Many people believe AI will eliminate software developers. I see the exact opposite,” he said. “AI isn’t taking developers’ jobs away. It’s enabling millions more people to turn their ideas into real products.”

It is an argument increasingly heard across startup ecosystems, including Southeast Asia, where engineering talent remains expensive, technical co-founders are hard to find, and many early-stage ideas never progress beyond slides or mock-ups. AI coding tools have changed what is possible at the prototype stage. The next challenge is whether they can also lower the cost and complexity of launching real services.

The hidden work behind every app

GENCOW’s starting point is not the visible side of software: slick interfaces, chatbots, dashboards or mobile screens. It is the infrastructure beneath them.

Before founding the company, Choi spent more than a decade and a half at Samsung Electronics, where he led the development of international software products and worked on large-scale launches. Over time, he noticed a pattern. As deadlines approached, backend engineers were often the ones working the latest nights.

That is because every new digital service requires far more than the feature a user sees. Teams must set up servers, databases, authentication, payments, APIs, cloud infrastructure, deployment systems and security controls. For AI products, there is another layer: connecting to models, managing data flows and ensuring the service can operate reliably outside a test environment.

“The polished applications users see are supported by countless hours of invisible engineering,” Choi said. “I watched talented colleagues spend nights and weekends handling repetitive infrastructure work.”

GENCOW was built around that pain point. Its platform provides common building blocks such as user authentication, database management, payment integration, AI connectivity and operational infrastructure. The idea is to let developers and founders focus on what makes their product distinct, instead of repeatedly rebuilding the same backend systems.

The company describes its approach as “Prompt to Production”, a phrase that captures a broader shift in software creation. Natural-language prompts can now produce code and functional prototypes. GENCOW wants to extend that process to services that can actually run in the market.

Why this matters in Southeast Asia

The timing is relevant for Southeast Asia’s startup market. Across Indonesia, Vietnam, the Philippines, Thailand, Malaysia and Singapore, founders are experimenting with AI products in education, logistics, finance, healthcare, agriculture and customer service. But many face the same constraint: it is easier to identify a problem than to assemble the technical team needed to solve it.

This is especially true outside major hubs such as Singapore, Jakarta, Ho Chi Minh City and Bangkok. A founder in agritech, for example, may understand crop supply chains deeply but lack access to engineers who can build a scalable platform. A teacher may know exactly where learning gaps exist but be unable to turn that insight into a usable AI tutoring product. Local operators often have strong domain knowledge, but software development costs can block them before they test demand.

That is where platforms like GENCOW could become relevant. If AI lowers the technical barrier to product creation, Southeast Asia may see more startups emerge from industry practitioners rather than only from traditional software teams.

Choi sees this as a redefinition of who gets to be a developer.

“In the future, being a developer won’t be limited to people with computer science degrees,” he said. “Entrepreneurs, designers, marketers, researchers, educators, anyone with expertise in solving real-world problems will be able to build software with AI.”

Also Read: 5 Seoul startups made their Southeast Asia debut at Echelon Singapore 2026 under the SBA pavilion

The claim should not be overstated. Production software still requires judgement, security awareness, product thinking and operational discipline. A badly designed fintech or healthtech tool can do real harm. But the direction of travel is clear: the early stages of software creation are becoming more accessible.

From coding assistance to company creation

GENCOW is not alone in chasing this opportunity. Globally, the market includes infrastructure and app development tools such as Google’s Firebase, AWS Amplify, Supabase, Vercel, Replit, Bolt and Lovable, each attacking different parts of the software-building workflow. Some focus on backend infrastructure, others on AI-assisted coding or front-end app generation. GENCOW’s challenge will be to show that its combination of AI service development and production infrastructure offers enough value in a crowded field.

For founders, the difference between these tools matters. A prototype builder helps create a working demo. A backend-as-a-service platform removes some infrastructure work. A deployment platform helps teams ship and scale. The next generation of AI development platforms is trying to combine these steps into a more continuous workflow, reducing the handoff between idea, code, backend setup and live product.

GENCOW has already found one route into the market through South Korea’s government-backed “Startup for Everyone” initiative, where it was selected as an official AI solution provider. Through the programme, the company works with aspiring entrepreneurs and early-stage startups building AI-powered services.

Choi said the ideas he sees range from agriculture and education to local community problems. In the past, many such concepts would have struggled to move forward because hiring developers was too expensive or difficult. Now, he argues, founders can test ideas faster and with fewer resources.

“In the past, building a new service often required months of development,” he said. “Today, with AI, teams can build prototypes in days, validate ideas quickly, and iterate much faster.”

The future developer may not look like one

The biggest question hanging over AI development tools is whether they reduce the need for engineers or simply change what engineers do.

Choi is firmly in the second camp. His argument is that developers will spend less time assembling routine infrastructure and more time solving harder problems: architecture, security, product quality, data governance and user experience. In other words, AI may not remove technical work, but it could push human effort higher up the value chain.

That matters in markets where engineering teams are stretched thin. A small startup in Southeast Asia rarely has the luxury of dedicated backend, DevOps, security and AI infrastructure specialists. If common technical work can be automated or packaged, lean teams can attempt products that previously required larger budgets.

There is also a human dimension to Choi’s thesis. He frames GENCOW not only as a productivity tool, but as a way to reduce the late-night burden on developers.

“I want software developers to spend less time on repetitive infrastructure work and more time solving meaningful problems,” he said. “I want them to leave the office earlier, have dinner with their families, and focus on innovation instead of rebuilding the same backend systems over and over again.”

That may sound idealistic in an industry known for tight deadlines and compressed launch cycles. But it points to a real shift. If AI can absorb more of the repetitive work, software creation could become less about who can grind through infrastructure fastest and more about who understands the problem best.

Also Read: Korea’s startup ecosystem is training founders, not just funding them

For Southeast Asia, where the next wave of digital products will need to solve local, fragmented and often offline problems, that shift could be significant. The region does not just need more apps. It needs more people with direct knowledge of real-world problems to have a practical path to building them.

GENCOW’s bet is that AI will make that possible — not by replacing developers, but by multiplying the number of people who can create software at all.

The post From Samsung to startups: Kevin Choi’s bet on AI-powered software creation appeared first on e27.

Malaysia’s push to build more home-grown technology champions is gaining a fresh injection of capital, this time across three very different but strategically important areas: warehouse automation, medical technology commercialisation, and assistive digital tools for the visually impaired.

VentureTECH, a Malaysian government-backed impact investment company, has invested US$7 million in three Bumiputera-led high-growth, high-value companies: Move Robotic, Recove Group, and Edote.

The investments come as Malaysia tries to deepen local participation in advanced industries, not merely as a consumer of imported technologies but as a builder of solutions that can serve domestic needs and eventually regional markets. For Southeast Asia, where many countries face similar challenges around industrial productivity, healthcare access, and digital inclusion, the bet is also about whether locally developed technologies can scale beyond national borders.

Also Read: Malaysia’s unicorn hunters: The startups raising millions and scaling fast

VentureTECH said the three companies were selected for their differentiated technologies, scalable business models, and commercial potential. The broader aim is to support Bumiputera companies that can contribute to Malaysia’s industrial competitiveness while delivering measurable socio-economic impact.

“At VentureTECH, we invest in companies with innovative technologies that solve real-world challenges, strengthen strategic industries and have the potential to scale sustainably,” said VentureTECH CEO Azizan Jaafar. He added that Move Robotic, Recove, and Edote represent the kind of companies the firm wants to back: commercially promising businesses that also address national development priorities.

The capital is also aligned with Malaysia’s Bumiputera Economic Transformation Plan 2035, or PuTERA35, which seeks to strengthen Bumiputera participation in higher-value economic activities.

Making automation easier to adopt

Among the three investees, Move Robotic addresses a pain point that is becoming more urgent across Southeast Asia: how businesses can automate warehouses and logistics operations without taking on large upfront costs or stitching together disconnected systems.

The company provides a warehouse and logistics automation ecosystem that combines Autonomous Mobile Robots, autonomous material-handling systems, Automated Storage and Retrieval Systems, pallet shuttle systems, and software tools such as its Warehouse Management System and Robot Fleet Manager.

In simple terms, this means Move Robotic is not just selling robots. It is trying to provide the hardware and software layer needed to help warehouses move, store, track, and retrieve goods more efficiently.

That matters in a region where e-commerce, third-party logistics, electronics manufacturing, and cold-chain distribution are all expanding. Many operators want to automate but face barriers such as high capital expenditure, uncertain integration timelines, and a shortage of in-house robotics expertise.

Move Robotic’s robot-as-a-service model is designed to reduce that friction. Instead of buying equipment outright, customers can adopt automation through a bundled subscription that includes implementation and ongoing use. The company aims to become Malaysia’s first provider of a fully bundled Robot-as-a-Service solution.

The model mirrors a wider shift in industrial automation globally, where manufacturers and logistics operators increasingly prefer usage-based or service-based systems rather than heavy one-off purchases. For Malaysia, having a local provider could also reduce dependence on foreign vendors and make customisation easier for domestic industries.

Turning medical research into market-ready products

Recove Group operates in a different but equally difficult part of the innovation chain: bringing healthcare research out of universities and laboratories and into hospitals, clinics, and patient care.

The company describes itself as a medtech commercialisation platform. It works with universities, researchers, and medical specialists to develop research-driven healthcare innovations into market-ready medical technologies.

Also Read: “Data, not hardware, is the real bottleneck in humanoids”: Matrix Robotics CEO Allen Zhang

This is a familiar bottleneck across Southeast Asia. Universities and hospitals often produce promising medical research, but many inventions fail to reach the market because researchers lack access to regulatory expertise, product development support, manufacturing partners, distribution channels, or commercial capital.

Recove’s role is to close that gap. One of its commercialised products is NEORUBIN, a non-invasive newborn jaundice screening device. Jaundice is common among newborns, and early screening is important to prevent complications. Non-invasive tools can help reduce discomfort for infants while allowing healthcare workers to screen more efficiently.

With VentureTECH’s investment, Recove plans to accelerate product commercialisation and advance additional medical technologies in its pipeline. If successful, the company could become part of a broader effort to build a stronger medical device industry in Malaysia, where healthcare demand is rising alongside an ageing population, higher chronic disease burden, and growing expectations for affordable care.

For Southeast Asia, medtech commercialisation has regional relevance. Many markets in the region face similar healthcare access issues, especially outside major cities. Locally developed devices that are affordable, practical, and designed around regional clinical workflows may stand a better chance of adoption than imported solutions built for very different health systems.

Digital inclusion for the visually impaired

The third company, Edote, focuses on assistive technology for the visually impaired community. Its flagship product, eBrelle, is described as the first standalone Braille laptop in Malaysia and Southeast Asia.

The device allows visually impaired users to access, create, and interact with digital content independently. Edote has integrated proprietary hardware, software, and accessibility-focused applications into a single platform.

The problem it is trying to solve is both social and economic. Digital tools are now central to education, employment, government services, and everyday communication. Yet many visually impaired people still face barriers in accessing mainstream technology, particularly when devices are expensive, poorly localised, or require multiple add-ons to function effectively.

By developing a standalone Braille laptop, Edote is attempting to make digital participation more practical and self-directed. VentureTECH said its support will help the company expand manufacturing capabilities, strengthen its product portfolio, and reach more users in Malaysia and across the region.

The opportunity is not limited to Malaysia. Across Southeast Asia, assistive technology remains underdeveloped, despite large populations of people with disabilities and rising policy attention on inclusion. If Edote can balance affordability, durability, localisation, and distribution, it could address a market that has often been overlooked by mainstream consumer technology companies.

Capital with a policy purpose

VentureTECH’s investment is not a conventional venture capital bet based purely on financial upside. As a government-backed impact investor, its mandate includes strengthening strategic industries and supporting companies that can deliver broader economic value.

That dual mandate is increasingly visible across Southeast Asia, where governments are using capital, procurement, grants, and policy frameworks to nudge domestic companies into higher-value technology sectors. The challenge is ensuring that such support produces companies that can compete commercially, not just survive through policy protection.

Also Read: China builds robot armies while the West chases robot brains

For Move Robotic, Recove, and Edote, the test will be execution: converting capital into stronger products, wider adoption, and regional market access. Each operates in a sector with clear demand, but also with long sales cycles, integration challenges, regulatory hurdles, or affordability constraints.

Azizan said VentureTECH’s role goes beyond providing funding. The firm works with investee companies to strengthen capabilities, unlock market opportunities, and scale sustainably.

That hands-on support could prove important. Malaysia has no shortage of technical talent or research output, but turning innovation into exportable companies remains a long-term task. By backing three Bumiputera-led firms in sectors tied to industrial productivity, healthcare, and inclusion, VentureTECH is placing a targeted wager: that the next generation of Malaysian technology companies can be both commercially viable and socially useful.

Whether these companies can move from national promise to regional relevance will determine the true impact of the investment.

The post Malaysia’s VentureTECH backs Move Robotic, Recove and Edote with US$7M appeared first on e27.

Oracle NetSuite’s SuiteWorld 2025 in Las Vegas opened with a slogan built for a keynote stage: “No Limits.” But strip away the staging and the roughly 8,000 attendees packed into the venue, and what emerged from three days of product reveals and executive interviews was a more measured story — one about data plumbing, conversational interfaces, and the uneven pace of digital adoption across emerging markets, including Southeast Asia.

e27 sat down with NetSuite Founder and Executive Vice President Evan Goldberg and Asia Head Amit Suxena on the sidelines of the conference. Between the two conversations and the keynote itself, three themes stood out, each with implications for founders and operators far beyond Las Vegas.

Learning one: AI is only as good as the data beneath it

The loudest applause line of Goldberg’s keynote wasn’t about a flashy new feature. It was a claim about architecture. NetSuite, he argued, was built 27 years ago around transactions — sales, orders, payments — rather than accounting entries after the fact.

Also Read: ChatGPT, commerce, and cloud: How NetSuite sees the future of work and business

“Transactions are the atomic unit of business,” Goldberg said, describing why this matters for AI. “AI doesn’t just guess; it starts from the source because your data is unified. AI understands your full context… it sees deep into your business, not just the surface.”

This is a subtler argument than it first appears. The enterprise software industry has spent the past two years bolting AI copilots onto existing systems. NetSuite’s pitch is that a copilot is only as useful as the data model sitting underneath it — and if that data lives in silos, spreadsheets, and reconciled-after-the-fact ledgers, AI is reduced to generating plausible-sounding guesses rather than grounded answers.

That’s the reasoning behind Ask Oracle, NetSuite’s conversational layer, which the company describes less as a chatbot and more as a connective tissue across a single data model. Responses come with drill-downs linking every number back to its source transaction, and “EXPLAIN” traces that show how an answer was derived, an attempt to make AI outputs auditable rather than opaque.

Suxena framed the shift in terms of what a finance leader can now do without touching a spreadsheet. “If you’re a CFO today, you no longer need to dig through ledgers or reconcile spreadsheets manually. You can simply ask, ‘where are my profits coming from?’, and the system gives you an intelligent, data-driven answer,” he said. “We’re moving from finance tools to business intelligence partners, from procurement systems to decision-support systems.”

The caveat, which both executives were careful to underline, is that none of this replaces judgement. “AI should never replace human judgment; it should augment it,” Suxena said. “AI tells you what’s happening and what could happen, but the why and what to do about it are still human calls.”

Learning two: Commerce is going conversational and nobody has fully worked out what that means for brands

The second theme to emerge was more speculative, and Goldberg was refreshingly candid about the uncertainty. Asked about e-commerce integration with ChatGPT, a feature NetSuite previewed during the keynote, he didn’t offer a tidy roadmap.

“No one has a clear answer to this question yet,” he admitted. “What’s clear is that commerce is moving into conversational interfaces.” He pointed to OpenAI’s “agentic commerce” push, which lets users complete purchases directly inside a chat window, as an early signal of where retail interactions might be heading.

Also Read: ‘AI sees deep into your business, not just the surface’: NetSuite’s Evan Goldberg

NetSuite is working with partners including Shopify to plug order flows into this emerging channel, but Goldberg was upfront about the tensions this creates. If a purchase happens entirely within a chatbot, how does a brand preserve its identity when the storefront — the visual language, the tone, the merchandising — is stripped out of the transaction? How does a smaller retailer differentiate itself when SEO and a distinctive website may carry less weight than they once did?

Goldberg brought the question home with a personal example: his wife runs a small business on NetSuite, and her chief worry mirrors the industry’s broader anxiety. “Her biggest concern is exactly that: ensuring her brand’s personality still comes through in how products are presented,” he said.

His answer, unsurprisingly, circles back to infrastructure rather than storefronts. Regardless of whether a sale originates in a browser or a chatbot, someone still has to manage inventory, coordinate shipping, and maintain real-time visibility across the order lifecycle. That backend discipline, he argued, is what will matter most as the front end of commerce keeps shifting.

Learning three: In SEA, the opportunity is timing, not budget

The third and most regionally relevant theme came from Suxena, who has spent years working with small and mid-sized businesses across the Philippines, Indonesia, and India. His observation cuts against a common assumption in enterprise software circles — that AI and cloud adoption in emerging Asian markets is primarily a cost or infrastructure problem.

“Many SMEs are still unsure where to start,” Suxena said. “They often ask, ‘Where should I use AI? I’m not in the business of technology.’ Our job is to take that uncertainty away.”

The bigger structural issue, in his view, is that founders treat digitalisation as a reward for reaching scale rather than a precondition for it. “Startups often make what I call ‘milestone decisions’ about digitalisation,” he explained. “They say, ‘When we hit 100 people, we’ll get an HR system,’ or ‘When we reach 500 customers, we’ll implement CRM.’ That’s backward.”

His advice, delivered without much room for nuance, was to digitalise from day one, not because scale demands it eventually, but because early automation compounds. “It’s not about when you reach those milestones; it’s about how fast you get there,” he said. “Waiting until you’re big enough often means you’re already struggling with inefficiency.”

Also Read: OpenAI calls for ‘AI infrastructure revolution’ to reboot Japan’s growth

Suxena also pointed to a geographic gap that’s easy to miss from Singapore or Jakarta, where SaaS adoption is now mainstream. The real headroom, he argued, sits in tier-two and tier-three cities across the region, where business owners are increasingly digital-native but haven’t yet been offered tools that are both accessible and affordably priced. “We’re seeing a democratisation of technology,” he said. “Cloud and AI are no longer just enterprise tools; they’re becoming everyday business essentials for the corner retailer, the logistics startup, the family-owned manufacturing firm.”

The common thread

Strip away the keynote choreography and NetSuite’s pitch at SuiteWorld 2025 rests on a fairly unglamorous premise: AI’s value is bounded by the quality and unity of the data it’s built on, whether that data sits inside a Las Vegas-built ERP system or a family manufacturing business in provincial Indonesia. The interfaces will keep evolving — conversational, agentic, embedded inside chat windows nobody had heard of two years ago. But for founders in Southeast Asia weighing when and how to invest in automation, the more durable lesson from this year’s SuiteWorld may simply be: start earlier than feels comfortable, and get the underlying data right before chasing the next AI feature.

The post Three key learnings from SuiteWorld 2025: Why context, not automation, is the real AI prize appeared first on e27.

As AI agents start searching for suppliers, negotiating terms, and initiating payments on a company’s behalf, the old assumption that a human approves every transaction is breaking down. A new report from Sunrate and Mastercard, Beyond Automation: Defining Agentic Global Payments, argues that speed alone will not make this shift viable — businesses will need what it calls a “Trust Layer”: infrastructure that verifies an agent’s identity, defines its authority, and records what it actually did.

The report frames this as fintech’s move from Know Your Customer to Know Your Agent, or KYA, built on three pillars: protecting credentials so agents never handle raw account details, capturing the intent behind a transaction, and enforcing governance that can authenticate an agent and revoke its permissions when needed.

The stakes are especially high for Southeast Asia, where startups juggle multiple currencies, jurisdictions and compliance regimes, and where, according to Gartner, half of all AI projects are abandoned after the pilot stage, often over unresolved risk controls. The report’s wager is that the winners in agentic payments will not be the startups with the flashiest AI interface, but those that can prove to banks, auditors and regulators that their agents’ actions were authorised, limited and traceable.

REGIONAL

SEA tech funding surges to 12-month high in July: Tracxn data shows the region’s startups closed 17 rounds last month, a 25.53% jump from June and 180.9% higher than a year earlier, as capital concentrated into fewer mega-deals.

Graas raises US$17M, buys Trustana for retail AI agents: The Singapore retailtech firm’s acquisition adds product-attribute and inventory data to its AI agents, aiming to close the gap between chatbots and SKU-level accuracy for online sellers.

K2 Therapeutics raises US$50M for global biotech push: The fresh funding positions the Singapore-based drug developer to compete internationally, addressing a gap in the city-state’s biotech scene: a steady pipeline of venture-backed drug developers built from day one.

Touchstone backs Vietnam’s N2TP AI research infrastructure: Unlike Vietnam’s earlier wave of consumer apps and fintech platforms, N2TP is building infrastructure for scientific research, where success is measured in hypotheses, experiments and patents rather than clicks.

Morph bets on stablecoins as the next rail for digital commerce: The Singapore firm argues stablecoins have outgrown their crypto-trading roots, offering freelancers and cross-border teams a faster, cheaper settlement rail than traditional banking channels.

What PayNow Gen 2 gets right, and what it risks under-valuing: The proposal to attach structured data to payments could let businesses reconcile transactions automatically, a back-office upgrade the columnist argues is more consequential than QR-code convenience.

Why Singapore’s AI finance race is now about data, not models: A new Forrester-cited analysis argues Singapore’s finance chiefs have moved past testing AI tools and now face a harder problem: making models work across messy, fragmented regional finance data.

Ryde taps HERE to improve ride-hailing routes and ETAs in Singapore: The tie-up with the mapping and location-data provider targets the “invisible layer” behind the app — traffic data, dispatch logic and arrival estimates — where rides are often won or lost.

Singapore’s next test in online child safety policy: As Australia, the EU and UK shift child protection into a political frame of its own, the columnist argues Singapore’s social media rules must evolve accordingly.

SEA’s biggest-market-first expansion logic is dead: Ranking markets by size and entering the largest first no longer works, the columnist argues, as founders increasingly need to sequence expansion by regulatory readiness and customer fit instead.

Vietnam’s born global startups are rewriting the playbook: Capital flowing into Vietnam-founded startups with global operations hit an all-time high in 2025, as founders increasingly skip the “win at home first” stage altogether.

Filipino virtual assistants are winning the remote-work race: Drawing on years in customer service and support, the writer argues the Philippines’ “I’ll try” work culture — not just cost — explains its outsized share of the global VA market.

Malaysia’s ZUS Coffee explores IPO to raise US$245M: Owner Zuspresso is working with financial advisers on a possible Bursa Malaysia listing that could value the coffee chain at around RM4 billion, with the offering potentially landing as soon as mid-2027.

Maybank: Shopee margins have bottomed, growth ahead: Sea Limited’s results reinforced the view that Shopee margins have bottomed, with management now expecting e-commerce adjusted EBITDA to surpass US$1 billion this year on rising ad take rates and VIP contribution.

Sea’s Q2 revenue jumps 48% to US$7.8B, income up 11%: Shopee and Monee drove the beat, with Shopee GMV up 28% and Monee’s loan book surging 62% to US$11.1 billion.

MSCI drops GoTo from Indonesia index over liquidity: The index provider cited low liquidity after GoTo’s shares stayed pinned near the exchange’s minimum tradable price for months, a technical call the company says is unrelated to its business performance.

U Mobile taps OpenAI to become an AI-enabled telco: The Malaysian telco gets early access to OpenAI’s frontier models across customer service, network operations and cybersecurity, marking OpenAI’s first telecoms partnership in Malaysia.

INTERVIEWS & FEATURES

ZERC’s founder on cracking SEA’s cooling crisis with paint: Parked cars in the region can hit cabin temperatures of 70-90°C within minutes — the radiative-cooling startup’s founder explains why a coat of paint, not air conditioning, is the fix.

The photographer who bet his business on AI, not against it: Rather than defend the craft against generative AI, SnappyFly founder Vincent Chow chose to drive the technology’s use in product photography himself, betting his Singapore firm on the shift.

INTERNATIONAL

Kospi enters bull market as crypto risk appetite fades: The S&P 500 and Nasdaq both rallied on AI-linked namessuch as CoreWeave and Super Micro, while Asian equities mirrored the optimism even as digital-asset markets lost momentum, the columnist notes.

Does US$1,780 support hold the key to an Ethereum rally: Ethereum developers are pushing an aggressive roadmap toward 10,000 transactions per second alongside quantum-safety upgrades, technical shifts the columnist ties directly to the coin’s next price move.

Is US$63,750 the only line between Bitcoin and US$62,000?: Total crypto market capitalisation fell 1.24% in 24 hours, a move the columnist reads as investors treating digital assets through a strictly macroeconomic lens rather than a technical correction.

5 US VC shifts every SEA founder should be tracking: While a Silicon Valley founder can get a term sheet within a week of a warm introduction, the piece argues SEA founders face a slower, more institutional fundraising process worth understanding.

Investors sue Selena Gomez over mental health startup: Plaintiffs who invested nearly US$1.2 million in Wondermind allege the singer and her mother committed securities fraud, claiming promised partnerships never materialised.

OpenAI hires Wiz’s Dali Rajic as CRO amid shake-up: Dali Rajic takes over sales after Denise Dresser’s nine-month tenure, the latest in a run of executive departures including COO Brad Lightcap and Fidji Simo.

Accel closes oversubscribed US$550M India fund in weeks: The firm still holds 55%+ of its previous fund undeployed, betting on AI applications, consumer internet, fintech and manufacturing.

Bluehill.VC closes maiden US$42M India deeptech fund: The Chennai firm’s debut fund will build a portfolio of 15-16 companies across defence, semiconductors and space.

CYBERSECURITY

The new ransomware playbook exposing ASEAN banks’ gaps: A regional bank learned of its own data breach from a regulator, not the attackers, a disclosure failure the piece says is becoming the norm across Southeast Asian financial institutions.

Education, energy, travel see rising cyber attack volumes: Organisations faced an average of 2,336 attacks per week in July, up 3% from June and 16% year-on-year, according to new Check Point Research data cited in the piece.

Crypto’s new threat is not a hack, but a knock at the door: A new Chainalysis report estimates more than US$30M lost to physical attacks on crypto holders, as Thailand and its neighbours become fronts in a distinctly offline category of crime.

Uber Freight probes breach claimed by hacking gang: The Helix group claims to have exfiltrated mailboxes, cloud storage and dispatch documents from the logistics subsidiary.

AI

Moving past the chatbox: agentic AI’s hidden enterprise risks: As APAC enterprises graduate from simple chatbots to fully autonomous AI agents, the piece warns that Model Context Protocol integrations bring governance and security risks many teams are unprepared for.

The hidden problem inside AI teams isn’t skills, it’s culture: An SME chief executive’s frustration over stalled AI adoption points to a workplace-culture gap that the writer says tools alone cannot fix.

From KYC to KYA: how AI agents are reshaping payment risk: As businesses deploy AI agents that can search, negotiate and pay suppliers autonomously, “know your agent” checks will soon matter as much as knowing your customer.

AI is not the beginning of drug discovery, it is the accelerator: Computational drug discovery predates generative AI by decades, tracing a lineage from molecular docking and QSAR modelling to today’s large models.

The language tax: Why AI skips your startup when buyers ask in Thai: AI assistants answer fluently in Thai, Vietnamese and Bahasa Indonesia but only recommend firms visible in those languages’ own sources, leaving startups invisible to their own regional buyers.

From chatbots to payment agents: AI’s next role in SEA commerce: A new report examines AI’s harder test in payments, where compliance failures are costly, once agents are plugged into real transaction workflows.

Sovereign AI starts long before the AI model: Infrastructure debates once about latency and cost are increasingly about where data can legally reside, shaping AI strategy before a single model is chosen.

The diagnosis is becoming free, the operation is not: AI will make differential diagnosis widely accessible, but the treatment that follows will remain expensive and unevenly distributed.

If AI can’t find your startup, does your startup exist?: Founders are increasingly asking ChatGPT, Gemini and Claude what they know about their own companies, a new front in startup PR.

The scarcity mindset is killing creativity, not AI: At Upscale Conf in San Francisco, speakers argued a scarcity mindset inherited from pre-AI resourcing constraints, not AI, is what’s limiting creative work.

The next AI payments boom may happen in the back office: The bigger AI shift is unfolding quietly inside finance teams, procurement departments and treasury desks few outsiders see.

How to choose the right AI marketing agency: Unlike traditional agencies, AI-driven ones work with real-time data and continuous experimentation, promising shorter cycles and lower costs.

SEMICONDUCTOR 

NXP breaks ground on expanded Malaysia chip factory: Dutch chipmaker NXP is building a900,000 sq ft facility in Malaysia, deepening Southeast Asia’s role in global semiconductorsupply chain diversification away from China.

Singapore chip parts maker UMS posts 90% Q2 profit jump: UMS Holdings beatexpectations on surging demand for semiconductor components, reflecting a broadrecovery in the chip equipment cycle that benefits Singapore’s precision engineering sector.

Nvidia’s US$500B manufacturing plan: risky but calculated: The chipmaker’s massivedomestic production bet hinges on repurposing ageing GPUs into a supply chain asset, amove with downstream implications for AI infrastructure buyers across Southeast Asia.

THOUGHT LEADERSHIP

How to make volunteer efforts actually sustainable: What starts as a simple conversation often spirals into paperwork, appeals and multi-agency coordination — a structural strain that burns out community volunteers.

Platform illusions: When ‘ecosystem’ is just a feature bundle: The word “ecosystem” appears in investor decks with near-magical confidence, even when what’s built is little more than bundled integrations and a marketplace tab.

You can’t force a tailwind, you can force your readiness for one: A decade into building, annual planning cycles stop being useful, since money arrives in bursts rather than evenly — making readiness the more valuable discipline.

The hidden cost of ‘gray work’ draining your team: Employees hired for one function increasingly find themselves buried in overlooked administrative tasks that compound quietly over time.

Good ideas are everywhere, venture capital isn’t: Airwallex’s decision to headquarter in Singapore rather than Melbourne was about more than location, since investors back the surrounding environment as much as the founders.

The system behind the smile: making volunteer efforts sustainable: Volunteer-driven community initiatives fail without structural support; the piece outlines frameworks for turning goodwill into durable, scalable operations.

The post Ecosystem Roundup: Why “Know Your Agent” will matter as much as KYC in payments appeared first on e27.

The post What PayNow Gen 2 gets right, and what it risks under-valuing appeared first on e27.

For retailers, artificial intelligence is only as useful as the data it can read. A chatbot that cannot understand product attributes, stock availability, customer intent, or the difference between two near-identical SKUs is unlikely to move the needle on sales. That is the gap Singapore-based Graas is trying to close with its latest fundraise and acquisition.

Graas, a retail commerce AI company, has raised US$17 million in a Series B round and acquired Trustana, a Singapore-based product-data platform. The round was led by LemmaTree, an investment firm founded by Singapore state investor Temasek, with participation from Integra Partners, Tin Men Capital, The Xander Group, IncredWealth and Orzon.

Also Read: AI shopping companions and the talent reset in retail

The financial terms of the Trustana acquisition were not disclosed.

Founded in 2022, Graas builds AI tools for retail commerce, drawing on more than US$1 billion in live commerce transactions processed through its platform.

Trustana, founded in 2020, helps retailers and distributors automate the enrichment and structuring of product information, the less glamorous but critical layer that determines whether a product can be discovered, compared, recommended, and sold accurately across channels.

Graas said Trustana’s product-data enrichment capabilities will be integrated into its Commerce Knowledge Graph, the underlying database that powers its AI agents for retail clients. In simple terms, a knowledge graph connects different pieces of information (products, customers, inventory, transactions, content, pricing, and behaviour) so that AI systems can make better decisions.

For a retailer, that could mean an AI agent which does not merely answer a customer’s question, but understands whether a product is in stock, whether a similar item has a higher margin, what the customer has bought before, and how to recommend the right option across chat, voice, or image-based search.

Why product data matters

The Trustana deal points to a wider shift in retail AI. Over the past year, much of the attention has gone to customer-facing generative AI tools: shopping assistants, automated ads, AI-written product descriptions, and conversational search. But these tools break down quickly when the underlying data is incomplete, inconsistent, or scattered across different systems.

This is a common problem in Southeast Asia, where retail is fragmented across online marketplaces, brand websites, social commerce, supermarkets, distributors, and general trade channels. A consumer may discover a product on TikTok, compare it on Shopee or Lazada, buy it through a brand site, and later repurchase it at a neighbourhood store. For large brands, stitching this journey together is difficult.

The challenge is even sharper for companies operating across multiple markets. Product catalogues are often maintained in different languages, formats, and systems. The same item can carry different descriptions, categories, pack sizes, and images depending on the sales channel. Without cleaning and structuring that data, AI agents risk producing poor recommendations or, worse, misleading answers.

This is where Trustana fits into Graas’ broader pitch. By adding Trustana’s product-data layer to customer and inventory information, Graas wants to build AI agents that can support both e-commerce and offline retail channels.

Prem Bhatia, co-founder and CEO of Graas, said bringing product, customer, and inventory data into a single system would allow AI agents to drive sales in both e-commerce and general trade.

Also Read: More choices, less hassle: Unlocking retail magic with AI and tech

That focus on general trade is significant in Southeast Asia. Despite the rise of digital commerce, a large share of consumer goods sales in the region still happens through small shops, distributors, and offline channels. For global brands such as Unilever or Puma, AI tools that only optimise marketplace ads or online storefronts solve only part of the problem.

A broader customer base

The acquisition also gives Graas access to Trustana’s customer relationships. Trustana’s clients include David Jones, Chemist Warehouse, and Toys”R”Us. Graas already serves brands including Unilever, Puma, and Schneider Electric, with operations across Australia, Southeast Asia, and the Gulf.

This regional spread matters. Retailers in mature markets such as Australia may have more structured digital operations, while Southeast Asian markets tend to involve more fragmented distribution and marketplace-led commerce. The Gulf, meanwhile, has seen rising investment in retail digitisation, particularly among large consumer brands and mall-based retail groups.

For Graas, combining these markets could offer a wider base of commerce data and use cases. The company says its platform already draws on more than US$1 billion in live commerce transactions, a figure that gives its AI systems more context on how consumers browse, compare, and buy.

Rebecca Xing, CEO of Trustana, said the combination would help the company accelerate growth and deliver more to its global customers.

Glenn Gore, CEO of LemmaTree, said the combined platform is positioned to help retailers make use of agentic AI — a term used to describe AI systems that can take actions on behalf of users, rather than simply generate text or analysis.

The term is becoming common in enterprise software, though it remains loosely defined. In retail, an agentic AI system could monitor inventory, recommend pricing changes, generate product content, suggest campaign adjustments, or respond to customer queries with purchase options. The practical value depends on how well the system is connected to real-time business data.

The competitive field

Graas is operating in a crowded and fast-changing market. Globally, large software providers such as Salesforce, Adobe, Shopify, Bloomreach, Algolia, and Dynamic Yield are embedding AI deeper into commerce, personalisation, search, and customer engagement tools. In Asia, companies such as Insider also compete in customer experience and marketing automation, while marketplace operators and e-commerce enablers offer their own analytics and optimisation layers.

Graas’ differentiation appears to lie in its focus on retail commerce data across multiple channels, rather than only storefront software or marketing automation. The Trustana acquisition strengthens that positioning by adding product intelligence, a foundational layer that many AI commerce tools still depend on but do not always control directly.

Still, execution will be the test. Enterprise retail customers are often slow to replace core systems, and many already use a mix of ERP, CRM, marketplace, and analytics platforms. To win larger accounts, Graas will need to show that its AI agents can plug into existing workflows, improve sales or margins, and reduce the manual work that typically sits behind catalogue management and commerce operations.

Singapore’s AI commerce play

The deal also reflects Singapore’s continued role as a base for regional enterprise technology companies. The city-state’s proximity to Southeast Asian markets, access to capital, and concentration of regional headquarters make it a natural launchpad for companies selling to large brands across Asia-Pacific and the Middle East.

For investors, retail AI offers a large but demanding opportunity. Consumer brands are under pressure to grow across more channels while managing tighter margins and higher customer expectations. AI promises efficiency, but retailers are increasingly looking beyond experiments and pilots. They want systems that can produce measurable commercial outcomes.

Also Read: Revolutionising retail: A blueprint for future success

Graas’ fundraise and acquisition suggest a bet that the next phase of retail AI will not be won by flashy interfaces alone. It will depend on whether companies can organise messy commerce data well enough for AI agents to act on it reliably.

In a region where retail remains both deeply digital and stubbornly offline, that may be the harder problem — and the more valuable one to solve.

The post Graas raises US$17M, acquires Trustana to build smarter retail AI agents appeared first on e27.

Cybersecurity teams entered the second half of 2026 with little sign of relief. Global organisations faced an average of 2,336 cyber attacks per week in July, up 3 per cent from June and 16 per cent from a year earlier, according to new data from Check Point Research.

The figures point to a threat environment that is not merely growing in volume, but widening in scope. Traditional attack channels such as email remain persistent, ransomware groups appear to have regained momentum, and generative AI tools are creating fresh data leakage risks inside companies faster than many security policies can catch up.

Also Read: Southeast Asia’s cyber boom is fuelled by fear—and AI

For Southeast Asian startups and digital businesses, the findings land at an uncomfortable moment. The region’s companies are adopting AI, cloud software and cross-border digital operations at speed, often with lean security teams and fragmented tooling. That combination can create the kind of gaps attackers look for: exposed credentials, unmonitored data flows, vulnerable suppliers and employees using new tools before governance catches up.

“July’s data shows that cyber risk is accumulating across multiple fronts at once,” said Omer Dembinsky, Data Research Manager at Check Point Research. “Attack volumes continue to rise, ransomware has accelerated sharply, and GenAI exposure is now part of daily business activity.”

Education and government remain in the firing line

The education sector remained the most attacked globally, with an average of 4,848 weekly attacks per organisation in July, a 14 per cent increase year on year. Government followed with 3,044 attacks, while telecommunications recorded 2,927.

The prominence of education is not surprising. Schools, universities and training institutions often hold large volumes of personal data, run sprawling IT networks, and operate with uneven cybersecurity budgets. In Southeast Asia, where governments have pushed digital learning platforms, online admissions and student management systems, these institutions can be attractive targets for both data theft and disruption.

Energy and utilities also saw a sharp increase, rising 20 per cent year on year to 2,759 weekly attacks per organisation. Hospitality, travel and recreation entered the top five with 2,614 attacks, up 28 per cent.

That matters for Southeast Asia, where tourism has rebounded strongly since the pandemic and travel operators have become deeply dependent on digital booking, payment and identity systems. A breach in this sector can quickly spill across customers, payment providers, loyalty programmes and third-party booking platforms.

APAC remains one of the world’s most attacked regions

Latin America recorded the highest attack volume in July, with 3,561 weekly attacks per organisation, up 19 per cent year on year. Asia Pacific followed closely at 3,316 attacks, ahead of Africa at 3,237.

Europe stood out for the pace of increase, rising 18 per cent year on year to 2,051 attacks per organisation. North America saw a 9 per cent increase to 1,613.

For Southeast Asia, APAC’s high ranking reflects a familiar structural issue. The region is home to fast-growing digital economies, but cybersecurity maturity varies widely between markets and sectors. A fintech in Singapore, an e-commerce platform in Indonesia, a logistics company in Vietnam and a hospital network in the Philippines may all be part of the same digital supply chain, but operate under different standards, budgets and regulatory pressures.

Also Read: What AI safety researchers actually worry about

This unevenness is a particular concern for startups. Many young companies rely on cloud platforms, software-as-a-service tools and outsourced development teams from day one. These choices help them scale quickly, but also widen the attack surface if access controls, vendor reviews and incident response plans are treated as later-stage concerns.

GenAI turns into a daily data risk

Perhaps the most modern risk in Check Point’s July data concerns generative AI. The research found that one in every 36 enterprise prompts carried a high risk of sensitive data leakage. Among organisations that regularly use GenAI, 88 per cent were affected by high-risk prompt activity.

The issue is not simply that employees are experimenting with chatbots. It is what they are putting into them. Check Point found that 22 per cent of prompts contained potentially sensitive information. Personal data appeared in 70 per cent of affected organisations, while financial data and network or IT infrastructure information each appeared in 68 per cent.

On average, organisations used eight GenAI tools, with users generating 95 prompts. In practical terms, this means employees may be feeding customer records, internal financial details, source code, contracts, credentials or system architecture into tools that were not approved or monitored by security teams.

For Southeast Asian startups, the risk is acute because GenAI has moved quickly from novelty to workflow. Founders use it to draft investor updates, developers use it to debug code, sales teams use it to summarise customer calls, and operations teams use it to process documents. Without clear rules, a productivity tool can become an unintentional data export channel.

The challenge is to govern AI use without blocking it outright. Companies will need policies that define what can and cannot be entered into public tools, technical controls to detect sensitive data in prompts, and safer enterprise-grade AI environments for teams that need to work with confidential information.

Email remains the old reliable route for attackers

Even as AI creates new risks, email continues to do what it has always done for attackers: provide a cheap, scalable entry point.

Check Point found that one in every 128 emails, or 0.78 per cent, was classified as phishing in July. Another 20 per cent fell into unwanted or risky categories such as graymail, spam and suspicious messages. Africa had the highest phishing rate, at one in every 106 emails, followed by North America at one in every 117.

Phishing remains effective because it targets people rather than systems. A single fake invoice, delivery notice, password reset request or investor email can be enough to trigger credential theft, malware installation or business email compromise.

In Southeast Asia, where companies often work across languages, currencies and jurisdictions, the room for deception is wide. A fraudulent supplier email or payment instruction can be difficult to spot when teams are already managing regional vendors, remote staff and multiple messaging channels.

Ransomware breaks from its earlier pattern

The clearest shift in July was ransomware. Reported ransomware attacks reached 964, up 49 per cent from June and 87 per cent compared with July 2025. That marked a break from the first half of 2026, when monthly activity averaged around 672 incidents.

Business services accounted for 32.5 per cent of reported victims, followed by industrial manufacturing at 14.4 per cent and consumer goods and services at 13.4 per cent. North America remained the most affected region, accounting for 45 per cent of incidents, while Europe followed at 28 per cent and APAC at 17 per cent.

The United States dominated the country-level victim count with 39.4 per cent of reported attacks, ahead of Germany, Canada, the United Kingdom and Italy.

Ransomware data based on published victims can undercount the real scale of incidents, as not every attack is disclosed or listed by criminal groups. Still, the July jump suggests attackers are finding enough success to sustain and expand operations.

The most active groups in July were The Gentlemen and Qilin, each responsible for 14 per cent of published attacks. DeadLock followed with 10 per cent and 97 reported victims, underlining how fluid the ransomware ecosystem remains as groups rebrand, fragment or compete for targets.

Also Read: Thailand is suddenly on the frontline of a new ransomware wave

For founders and operators, the takeaway is blunt: cybersecurity is no longer just an enterprise IT problem. It is a business continuity issue. As attack volumes rise and AI reshapes both productivity and exposure, companies that treat security as an afterthought may find that the cost of catching up arrives all at once.

The post Education, energy and travel sectors face rising cyber attack volumes appeared first on e27.

Few words in business are abused as generously as ecosystem.

It appears in investor decks, product strategy documents, market narratives, and executive speeches with almost magical confidence. A company launches adjacent tools, adds integrations, bundles workflows, introduces a marketplace tab, and suddenly begins speaking as though it has crossed into a higher strategic category. The implication is always the same. This is no longer just a product. This is now a platform. This is now an ecosystem. This is now a position of structural power.

Most of the time, it is not.

What many companies call an ecosystem is simply a feature bundle with better language around it. It is a larger product surface, not a different market role. It may be useful. It may even be commercially smart. But usefulness is not the same as platform power, and adjacency is not the same as ecosystem formation.

An ecosystem begins when value creation is no longer controlled only by your roadmap

This is the simplest way to separate platform reality from platform theatre.

If nearly all meaningful value still comes from what your internal teams choose to build, package, release, and sell, then you do not yet have an ecosystem. You have a company with an expanded product portfolio. That may be a good business. It is not the same thing.

An ecosystem starts when external actors begin creating value that is meaningful to customers and meaningful to themselves because your platform exists. That means partners, developers, service providers, data contributors, implementation specialists, operators, or adjacent businesses are not merely attaching themselves to your marketing story. They are making decisions, investments, and in some cases their own commercial bets based on your presence in the market.

That is where the strategic category changes. The centre of gravity shifts from what you build to what others can build, offer, sell, certify, customise, govern, or depend on because your system provides the base layer.

This is why so many claimed ecosystems are overstated. The company still owns the proposition, owns the commercial model, owns the roadmap, owns the customer relationship, and owns nearly every meaningful form of innovation. Everyone else is decorative. They may integrate, implement, or resell, but they are not genuinely extending the market in a way that creates new value beyond the vendor’s own design.

Also Read: Why money won’t save Bangladesh’s startups: The ecosystem readiness crisis

Feature breadth is not the same as market orchestration

Many firms confuse having more things with holding a more important position.

The logic usually sounds persuasive at first. We now offer workflow A, workflow B, analytics, automation, collaboration, reporting, and compliance in one environment. Customers use more of our modules. We have integrations with major third parties. We have a partner page. We are becoming the centre of the ecosystem.

Perhaps. But perhaps not.

Feature breadth tells you that the company is occupying more use cases. It does not tell you whether the market is beginning to organise around the company as a coordinating layer. Those are very different conditions.

The real test is whether others can build serious economic logic on top of you

Can another business create durable economics because your platform exists, without simply acting as your implementation arm or distribution helper?

Not can they list themselves in a marketplace. Not can they complete an API connection. Not can they appear in a partner brochure. Can they build real business logic around your platform?

Can they specialise around it? Can they innovate on top of it? Can they develop differentiated offers because of it? Can they make investments that make sense only if your platform continues to matter? Can they gain customers, revenue, data, reputation, or operating leverage through participation that is not entirely controlled by your next release cycle?

Real platforms create politics, fake ones create packaging

There is a harder truth here that many executives would rather avoid.

A genuine ecosystem is not just larger. It is more difficult to govern. Once external actors begin relying on your platform for their own outcomes, you no longer have the luxury of pure product thinking. You now have politics.

You have to decide who gets access and on what terms. You have to decide how disputes are handled. You have to decide whether the platform favours its own products over third parties. You have to decide how standards evolve, who bears integration cost, how abuse is controlled, what quality thresholds apply, how data rights work, and what happens when the platform’s own interests conflict with those of participants building on top of it.

Also Read: Southeast Asia’s investors are sleeping on a US$2B ecosystem next door

Many ecosystem claims are really attempts to borrow strategic prestige

Part of the reason this language spreads so easily is that platform sounds like a more powerful category than product.

A product sounds finite. A platform sounds expansive. A product competes on features. A platform shapes markets. A product can be replaced. A platform becomes infrastructure. Leaders know this, investors know this, and the language becomes attractive very quickly.

So firms start narrating themselves upwards.

A company with adjacent modules begins to speak as though it has become a platform. A firm with a few partners begins to imply network effects. A vendor with bundled workflows starts describing market orchestration. The ambition may be genuine, but the language often outruns the operating reality.

The hidden issue is whether the company is willing to surrender control

Many firms say they want ecosystem dynamics, but what they really want is ecosystem valuation without ecosystem loss of control. They want others to extend the product, increase reach, add use cases, and create market energy, while the company still decides everything that matters.

That tension usually sits at the heart of the illusion.

A real platform has to surrender something. It has to allow external actors enough room to create meaningful value. It has to tolerate less central control over the total experience. It has to accept that innovation, customer intimacy, and even some forms of commercial power will now exist outside its direct management. It has to govern rather than simply command.

A bundle can still be a very good strategy, it just is not the same strategy

There is nothing weak or unserious about building a tightly integrated product suite. In many markets, that is exactly the right move. Customers may prefer one accountable vendor, cleaner workflows, faster procurement, simpler support, and less complexity. A broader feature set can deepen retention, increase share of wallet, and improve strategic relevance without any need for ecosystem theatre.

The problem is not bundling. The problem is pretending bundling and platform formation are the same thing.

They are not.

A bundle strategy is about offering more direct value yourself. A platform strategy is about enabling and governing value creation by others. A bundle strategy can be highly profitable and highly defensible. But it should be understood honestly, because the operational demands, investment logic, partner model, governance model, and eventual sources of power are different.

The company that confuses these paths usually ends up doing both badly. It never fully commits to platform openness, yet it also stops treating product coherence as its true strategic centre. It starts speaking like an orchestrator while operating like a suite vendor. That gap becomes visible over time.

—

The post Platform illusions: When ‘ecosystem’ is just a feature bundle appeared first on e27.

For many finance chiefs, the first wave of AI was about testing tools: automating reports, speeding up reconciliation, or asking software to spot anomalies in spreadsheets. In Singapore, that phase is quickly giving way to a more difficult question: how to make AI work across the messy reality of regional finance operations.

A new Forrester Consulting study commissioned by Airwallex suggests Singapore is among the more advanced markets globally in operationalising AI within finance functions. But it also points to a constraint that will be familiar to many Southeast Asian companies expanding across borders: fragmented systems, inconsistent data, and legacy workflows are now bigger obstacles than access to AI models themselves.

Also Read: From KYC to KYA: how AI agents are reshaping payment risk

The study surveyed more than 1,200 finance decision-makers across 11 markets, including Singapore, and was launched as part of Airwallex’s Business Builders programme, a Singapore initiative featuring founders and finance leaders from companies such as Endowus, StaffAny, CardUp, GlobalTix, Peakflo, Polybee, Chronos Agency, Stryv and Pitstop Tyres.

The findings capture a shift in the region’s AI conversation. Adoption is no longer the headline. Execution is.

The AI budget is still rising

Singapore finance leaders are not pulling back from AI. According to the study, 96 per cent expect investment in AI-powered finance to increase over the next 12 months. That figure reflects how quickly AI has moved from a side experiment to a core operating priority.

In finance teams, AI is being used for tasks such as bookkeeping, reporting, forecasting, fraud detection, compliance checks and scenario modelling. These are areas where speed and accuracy matter, but where human teams are often slowed down by manual processes and scattered data.

The appeal is clear. In Southeast Asia, even relatively young companies often operate across multiple markets, currencies, banks, payment rails and regulatory regimes. A Singapore-headquartered SaaS, fintech, travel or e-commerce startup may collect revenue in Indonesia, pay vendors in Vietnam, hire teams in the Philippines and raise capital from overseas investors. That creates a level of financial complexity that spreadsheets and disconnected tools struggle to manage.

AI can help, but only if it can see the full picture.

That is where many finance teams are getting stuck.

Fragmented data is the real bottleneck

While AI adoption is high, scaling it remains harder. In Singapore, 64 per cent of finance leaders identified fragmented or inconsistent data across disconnected systems as a core barrier to scaling AI. Globally, 65 per cent cited the same issue.

This matters because AI systems depend on clean, timely and connected information. If transaction data sits in one system, procurement in another, payroll somewhere else, and regional subsidiaries use different reporting formats, AI can only produce partial insights. In some cases, it may automate bad assumptions faster.

The study also found that 68 per cent of Singapore respondents said their finance workflows are only partially digitalised, while 53 per cent said data either flows inconsistently across finance platforms or remains largely siloed.

For a region such as Southeast Asia, this is not a minor operational issue. Many companies expand market by market, often adding tools as they go. A payment provider may be chosen for one country, an accounting platform for another, and a separate expense tool for a newly opened office. What works in the early stages can become a constraint as the business grows.

Arnold Chan, General Manager for Asia Pacific at Airwallex, framed the challenge directly: “Businesses are no longer asking whether to invest in AI. They’re asking how to make AI work at scale.”

Also Read: The language tax: Why AI skips your startup when buyers ask in Thai

He added that the biggest obstacle is not access to AI models, but the financial systems beneath them. “Businesses that connect their financial data, workflows and operations will be far better positioned to move beyond isolated AI use cases towards more intelligent, autonomous finance.”

Singapore is ahead, but not fully autonomous

The study suggests Singapore finance teams are further along than their global peers in allowing AI to run parts of finance operations with limited human involvement.

Eighteen per cent of Singapore respondents said AI already runs autonomously with minimal human input across finance workflows, compared with 11 per cent globally. In record-to-report processes, which include bookkeeping, closing and reporting, 27 per cent of Singapore finance leaders said AI runs autonomously, compared with 14 per cent globally.

That does not mean finance departments are handing over decision-making wholesale. Much of the near-term opportunity remains practical rather than futuristic.

In Singapore, 64 per cent of respondents expect AI to generate cash-flow forecasts and what-if analyses that recommend actions for humans to decide on within the next year. Globally, the figure is 51 per cent. Another 43 per cent of Singapore finance leaders expect AI to identify patterns, trends and root causes while leaving decisions to people.

This distinction is important. In finance, especially in regulated sectors such as fintech, wealth management and payments, full automation carries risks. AI-generated forecasts may be useful, but companies still need accountability, audit trails and human judgement when decisions affect cash, compliance or customers.

For Southeast Asian startups, where capital efficiency has become a sharper priority since the funding slowdown, better forecasting can still be valuable. Knowing earlier when working capital will tighten, when supplier payments may clash with payroll, or when regional revenue is drifting from plan can give management teams more room to act.

Talent becomes part of the infrastructure

The study also points to a second layer of readiness: people.

Singapore appears ahead here too. Twenty-seven per cent of finance leaders said their organisations have implemented structured, enterprise-wide AI talent strategies covering role redesign, certifications and hiring, compared with 15 per cent globally. Another 25 per cent said they have in-house AI development capabilities within or closely aligned to finance, versus 16 per cent globally.

This is significant because AI in finance is not simply a technology upgrade. It changes how teams work. Finance professionals may need to understand how to validate AI outputs, design workflows, question recommendations and work with engineering or data teams. The role moves from compiling information to interpreting and governing it.

At the same time, the study suggests not every company wants to build everything internally. Sixty-six per cent of Singapore finance leaders expect to use a hybrid model over the next year, combining in-house expertise with external providers. The proportion planning to build AI entirely in-house is expected to fall from 32 per cent today to 17 per cent.

That reflects a pragmatic reality. Even well-funded companies may not want to maintain large internal AI teams for finance alone. The more likely model is a mix of finance platforms, internal data capability and external specialists.

What this means for Southeast Asian companies

The broader lesson is that AI advantage in finance may depend less on who adopts the newest tool and more on who fixes the foundations first.

For startups and growth companies in Southeast Asia, this can be uncomfortable. Infrastructure work rarely attracts the same attention as product launches or fundraising rounds. But connected finance systems can determine whether AI becomes useful in daily decision-making or remains trapped in isolated pilots.

Also Read: From chatbots to payment agents: AI’s next role in SEA commerce

Singapore’s position as a regional headquarters market gives it a natural lead. Many companies base finance, strategy and investor relations teams in the city-state while operating across the rest of Southeast Asia. That makes Singapore a testing ground for AI-enabled finance models that may later be applied across more fragmented regional markets.

The next year will show whether companies can turn AI investment into operational change. The money is flowing, and the tools are improving. The harder task is making sure the data, systems and people are ready for them.

The post Why Singapore’s AI finance race is now about data, not models appeared first on e27.