The post Bitcoin rejected US$84,000 and forced US$79,060,000 in long positions out appeared first on e27.

I’ve been running The Wellness Insider since 2016, and over the years, there have been countless occasions when I’ve looked at my website and thought, “I wish I could change that easily.”

Sometimes, it was something relatively minor, such as the spacing between menu items or the way an image appeared on a page. Other times, I wanted to change an entire section of the website, improve how readers navigated between articles or introduce new features.

The problem wasn’t that I didn’t know what I wanted. It was that I didn’t know how to make it happen.

I’m a marketer and content creator, not a web developer. I understand website structure, user experience and what makes a digital platform effective. I can navigate WordPress, publish articles and make basic changes with my limited html knowledge. But when something requires editing code or troubleshooting a technical issue, that’s usually where my abilities end.

For years, that meant either living with the limitations of my website or paying someone to make the changes.

Then I started using AI to help me do something that was beyond my technical capabilities: revamp my own websites.

Knowing what you want is different from knowing how to build it

The Wellness Insider is a digital wellness publication, so the website isn’t merely a blog or an online magazine. It’s where readers discover articles, browse events and engage with our content. It’s also part of how we work with advertisers and partners.

As the publication evolved, I became increasingly dissatisfied with certain aspects of the website.

I wanted a more polished homepage that would give greater prominence to our latest stories. I wanted readers to be able to discover more articles without navigating through unnecessarily complicated menus. I also wanted to improve our events section, making it easier for people to find upcoming activities as we have seen how some of our event listings are very popular with our readers.

These weren’t particularly revolutionary ideas. They were improvements that many established websites already had.

But knowing that something should be improved and knowing which WordPress settings, CSS rules or template files to modify are two very different things.

I could explain exactly what I wanted a page to look like. I just couldn’t translate those ideas into code.

This is where AI changed things.

From describing a problem to actually fixing it

I began using ChatGPT to first analyse my website and give me the blunt truth about what works and what could be improved. After which, I looked through its suggestions and picked what I liked about them while also describing what I disliked about my site and what I wanted instead.

Rather than searching through countless tutorials, trying to figure out which instructions applied to my particular WordPress theme, I could explain the issue in ordinary language.

Also Read: What does ChatGPT say about your brand? Miracly raises US$2M to find out

For example, I wanted The Wellness Insider’s homepage to look more like a premium digital publication. We worked on giving the logo greater prominence, featuring a selection of articles more effectively and improving the layout of the newsletter section.

I also wanted to improve our events calendar. What started as a request to make events easier to discover developed into a more substantial redesign, combining a calendar view with event listings and a submission form.

There were smaller improvements too: correcting menu spacing, fixing sidebar behaviour, adjusting article layouts and dealing with elements that weren’t displaying as intended in the original theme template.

Individually, some of these changes might seem insignificant. But when you operate a content-heavy website, small frustrations accumulate. A poorly positioned element, a confusing navigation menu or a sidebar that behaves strangely can affect how people experience your content.

What I appreciated most was that I could approach these issues from the perspective of a website owner rather than a developer.

I didn’t need to know the technical solution before asking for help. I needed to understand the problem well enough to describe it.

Then I tried it on another business

Encouraged by the improvements to The Wellness Insider, I began working on the Proving Grounds website.

Proving Grounds operates in the gaming and esports space, so its website serves a different purpose. It needs to communicate with the gaming community while also presenting the business professionally to potential corporate clients and partners.

Our previous website wasn’t doing both particularly well (sorry hubs!).

I wanted a design that reflected the brand’s personality, with a darker visual identity, stronger typography and clearer navigation. I also wanted the site to distinguish between community activities and the services we offer organisations.

This project involved a bigger change because we were moving away from a website-builder setup towards a more flexible WordPress site.

AI helped me work through the structure, page layouts, design requirements and technical implementation.

I could specify the brand colours and fonts, explain how I wanted different sections organised and review the results. When something wasn’t right, I could ask for it to be changed quickly.

We also worked on functionality, including navigation, enquiry forms and a separate mascot-naming campaign with a voting form.

It was particularly satisfying to see ideas that had previously existed only in my head become functioning website features.

I still wasn’t writing the code myself. But I was no longer entirely dependent on someone else to interpret and implement my ideas.

Also Read: Employers are buying AI tools. Their managers are being left behind

Of course, AI didn’t get everything right

I don’t want to give the impression that I typed a few instructions, went for coffee and returned to two perfectly redesigned websites.

There were plenty of moments when something didn’t work as expected.

A layout might look fine in one area but create problems elsewhere. A change could require additional adjustments. Sometimes, an issue that appeared to have been resolved needed another round of troubleshooting.

And there were occasions when I had to repeat instructions or remind AI of decisions we’d already made.

This was frustrating, particularly when I thought we had finished working on a particular section.

It also taught me something important: using AI doesn’t remove the need to understand what you’re trying to achieve.

I might not know how to write CSS or PHP from scratch, but I still need to recognise when a website looks wrong, when a feature isn’t behaving properly or when a proposed change doesn’t make sense for the business.

AI can help with implementation. It cannot take responsibility for the final result.

I also learnt to be more specific with my instructions. Saying “make this look better” leaves considerable room for interpretation. Explaining what feels wrong, what should change and what needs to remain untouched generally produces more useful results.

In that sense, working with AI isn’t entirely different from briefing a human developer or designer.

The clearer the brief, the better the chances of getting something usable.

The biggest benefit wasn’t just saving money

As a solopreneur, I’m constantly balancing what needs to be done against the time and resources available.

Hiring a developer for every small website adjustment isn’t always practical. Even when the cost is manageable, there’s the additional process of explaining the problem, requesting quotations, coordinating changes and checking the completed work.

AI has helped reduce some of that friction.

I can experiment with an idea without immediately committing to a development budget. I can make incremental improvements instead of waiting until I’ve accumulated enough changes to justify hiring someone.

Also Read: Tokens are the wrong yardstick for AI agents, says JFrog’s Sunny Rao

But the benefit goes beyond cost savings.

Previously, I sometimes left things unchanged because I didn’t have the technical knowledge to tackle them. Now, I’m more inclined to ask whether something can be improved and explore how to do it.

That’s a significant shift in mindset.

It doesn’t mean I suddenly possess the skills of an experienced developer. Nor does it mean I would use AI alone for every technical project, particularly one involving complex security, payment systems or sensitive customer information.

What it means is that the threshold for attempting improvements has become much lower.

For a small business owner, that matters.

AI hasn’t made me a developer. It has made me less constrained by what I don’t know

One of the most interesting things about this experience is how it has changed my relationship with technology.

I’ve always been comfortable with digital marketing tools, content management systems and website analytics. But coding felt like a separate discipline, one that required knowledge I didn’t possess.

AI hasn’t eliminated that knowledge gap. I still don’t fully understand every line of code involved in the changes we’ve made.

What it has done is give me a way to work across that gap.

I can bring my understanding of branding, content, audience behaviour and user experience to a technical project, while AI helps translate those requirements into something that can be implemented.

That doesn’t make human expertise irrelevant. Experienced developers bring architectural judgement, security knowledge and technical understanding that an AI-generated solution still needs to be assessed against.

But not every website improvement requires rebuilding an entire system or engaging a development agency.

Sometimes, a business owner simply wants to fix an awkward layout, improve navigation or introduce a feature that makes the website more useful.

For me, AI has made those improvements more accessible.

And perhaps that’s what I appreciate most about using it.

It’s not that AI knows everything, or that it always gets things right. It certainly doesn’t.

It’s that I can finally do something about the parts of my websites that have been bothering me for years, instead of looking at them and thinking, “I wish I knew how to fix that.”

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post I used AI to rebuild my websites without knowing how to code appeared first on e27.

Taiwan’s Fair Trade Commission has blocked Delivery Hero’s sale of foodpanda Taiwan to Grab, scuttling a US$600 million all-cash deal agreed in March. The regulator said the remedies on offer could not offset the harm to competition.

The sticking point is Uber. The US company holds about 13.1% of Grab, a legacy of their 2018 regional deal, and the commission argued the stake would blunt rivalry between a Grab-owned foodpanda and Uber Eats, even though Grab’s dual-class structure caps Uber’s votes at 3.9%. It also cited the 2018 merger itself, for which Singapore’s competition watchdog fined both companies about US$9.5 million. There is a further twist: Uber agreed in July to buy Delivery Hero, leaving it on every side of the table.

This is the second time in under two years that Taipei has stopped the sale, after it rejected Uber’s US$950 million bid in December 2024.

Foodpanda Taiwan generated US$1.8 billion in GMV last year, and the deal would have handed Grab just over half the market. Delivery Hero can appeal.

For Southeast Asia, the ruling is a preview: if Uber’s Delivery Hero deal closes, regulators across the region will face the same cap-table questions.

Editor.

REGIONAL

Singapore security robot maker Kabam Robotics to shut down: The startup gave customers no reason for the closure in its email. SingHealth has deployed its robots at hospital facilities, and it remains unclear what happens to existing deployments or staff; LinkedIn lists 54 people linked to the firm.

VinFast’s US$500M Philippine plant bets on exports, not local demand: Built for 200,000 vehicles a year, the plant could produce six times local EV demand forecast for 2026. Construction starts in Q1 2027, while VinFast’s Q2 loss from continuing operations widened 128% to US$1.2B.

Carro launches Singapore’s first car-loan securitisation at US$350M: The MAS-approved vehicle buys loans from subsidiary Genie Financial Services and lets multiple lenders join through notes or direct loans. Carro, which transacted over 120,000 vehicles across eight markets in FY2026, wants to diversify funding for regional expansion.

Malaysia’s US$110M startup push runs into a deal-matching problem: Budget 2027 tasks Khazanah’s Jelawang Capital and KWAP’s Dana Perintis with deploying US$110.4M into startups. Investors say money is not the bottleneck; finding everything meant to accompany it is the harder part.

Ex-PropertyGuru CEO’s Genie Health raises US$3M to unify health data: The Singapore startup’s seed round, led by the Nadathur family office, funds an app that pulls screening reports and wearables into one view, targeting health data now scattered across PDFs, smartwatches and clinic records.

Homage turns in-house care software into a product as Singapore ages: The home care company is commercialising the rostering and visit-note tools built for its own care managers, betting Singapore’s super-aged shift will lift demand for back-office automation across providers.

Thunes opens a real-time payout corridor into Venezuela: The Singapore payments firm now routes instant transfers through Venezuela’s interbank platform, targeting a market where remittances hit about US$6B in 2025, roughly 7% of GDP, half of it sent from the US.

UK and Singapore widen cooperation on tokenisation, AI and quantum: The 11th UK-Singapore Financial Dialogue covered Project Guardian, stablecoins and MAS’s SAFR framework for AI agents handling financial tasks, alongside post-quantum migration and transition finance.

Circle says stablecoins fit Vietnam’s cross-border leg, not checkouts: With 17-21 million Vietnamese holding digital assets, Circle argues USDC could cut pre-funding delays in remittance settlement while recipients still receive dong, as Vietnamese law bars digital assets as payment.

Sunway lines up WeRide, Neolix and AutoFlight trials in Malaysia: The transport ministry has conditionally approved a pilot of two autonomous buses on BRT-Sunway Line. The driverless delivery and eVTOL plans still await technical testing and regulatory clearance.

Taiwan brings 33 startups to SWITCH 2026 in Singapore: Led by the National Development Council-backed Startup Island TAIWAN, the delegation exhibits at Sands Expo from 27 to 29 October as part of Taiwan Tech Week in Singapore.

INTERVIEWS & FEATURES

SEA’s US$1T digital economy target slips to 2035 as AI props up hope: This year’s e-Conomy SEA report projects US$368B in GMV for 2026, up 17%, but pushes the trillion-dollar milestone back five years, with AI now carrying the region’s optimism.

Global founders turn inward while SEA entrepreneurs look abroad: HSBC’s Global Entrepreneurial Wealth Report 2026 finds the share eyeing international expansion fell six percentage points globally, while Southeast Asian business owners buck the inward turn.

HK and Taiwan founders ride the AI boom; Southeast Asia misses out: A year after ranking among HSBC’s most cautious respondents, founders in Hong Kong, Taipei and Shanghai are now the most upbeat in HSBC’s survey, a turnaround Southeast Asian peers have yet to share.

SEA merchants must now optimise their listings for AI shopping agents: Shopee went live inside ChatGPT across seven markets in June and Lazada runs its Qwen-powered agent Lazzie, so sellers who optimised for human shoppers must now win over machine buyers too.

Techable’s Amulya Vishwanath: physical AI is still in its LEGO era: The ex-Nvidia marketer turned VC says robotics’ ChatGPT moment awaits foundational hardware and software building blocks such as actuators and hands, and calls talent access her firm’s moat.

INTERNATIONAL

Bain Capital commits up to US$250M to Hong Kong’s FundPark: The package starts with US$100M upfront, with US$150M more subject to conditions. FundPark has lent US$10B cumulatively to over 40,000 online sellers across Asia, underwriting with real-time sales data.

Walmart welcomes AI shopping agents as Amazon shuts them out: Its Agent Gateway MCP server lets approved agents search and buy, while Agent Profiles let shoppers cap each agent’s spending limit and approve transactions. Non-compliant agents get blocked.

Meta’s Muse passes 10M users, with shopping a top use: About 25% of users open the AI agent daily, and over a quarter of engaged shoppers average more than US$200 per purchase. Meta expects Muse to eventually out-earn its social apps per user.

Tesla to sell driverless Cybercabs by year-end despite NHTSA probe: Fleet buyers could run the steering wheel-free two-seater before unsupervised FSD reaches owners in Q2 2027, but regulators are still examining whether Tesla’s safety self-certification holds.

CYBERSECURITY

AI agents hacked 100 firms at US$50 each; SEA’s shops fit the profile: One operator using the Cairn agent platform targeted mid-sized online shops for payment-system access. With only 68.5% of ASEAN SMEs running specialised security software, the region’s D2C brands fit the profile.

New Mexico court orders Meta to pay US$25B over privacy deception: A jury found 43 million consumer-protection violations, each carrying a US$5,000 penalty, partly tied to the Cambridge Analytica-linked app. Meta calls the ruling a tax on speech and will appeal.

Scouter brings a listen-first pitch to Singapore’s busy cyber market: Scouter pitches diagnosing security teams’ real problems before selling products, a rebuke of the slide-deck-and-quarterly-target routine that security leaders at large Singapore organisations know too well.

ChatGPT adds PIN-locked chats, but no end-to-end encryption: Locked chats sit in a separate section with their own memory, secured by PIN or biometrics. OpenAI says the feature does not change data retention, and resetting the PIN deletes everything locked.

SEMICONDUCTOR

BMI: AI capex will keep driving Asia Pacific growth into 2027: Taiwan, South Korea and Singapore gain most from chip and electronics demand, while Malaysia draws data centre investment. BMI flags a hyperscaler slowdown and trade restrictions as key risks.

Chinese chipmaker CXMT’s shares slide 7.8% despite swing to profit: The DRAM maker forecasts January–September net profit of up to US$19.8 billion, reversing a US$790 million loss, as AI drives shortages. Its global DRAM share doubled to 10%, but investors fear a supply glut.

Qualcomm to cut jobs worldwide in performance-focused layoffs: CEO Cristiano Amon’s memo says notices begin in November but gives no headcount; Qualcomm had about 52,000 staff in 2025. Weak handset demand and shrinking Apple revenue weigh on the chip designer, which bets on data centres.

AI

DeepSeek reportedly raises US$15B days after US federal ban: Bloomberg says CATL and Tencent are among potential investors in one of China’s largest rounds, following a US$7B raise in June at a US$50B valuation. A Chinese IPO could follow next year.

GoTo’s 27B model speaks Batak, but who can afford to run it?: The model targets regional languages such as Batak Toba that global chatbots handle poorly, but the cost of running it may decide whether smaller Indonesian firms can actually use it.

Enterprise Singapore commits US$35.2M to push AI into SME factories: The three-year Catalytic AI Solutions scheme pairs SMEs with vendors and trade bodies to deploy proven tools for design, quoting, inspection, energy monitoring and predictive maintenance.

Singapore launches embodied AI challenge built on a Micron use case: Teams worldwide will program robots to service a mock semiconductor chamber, competing for a US$69,000 prize pool. Simulation runs to 31 December, with finals at RoboSG! 2027 in March.

Agentic AI tops Singapore’s job-skill demand, Jobstreet data shows: Jobstreet by SEEK’s AI Gauge shows recruiters rewriting job ads around agentic AI workflows, which raises hard questions for workers whose routine tasks those agents are built to absorb.

Enterprises can’t say who settles disputes between their AI agents: An OutSystems report finds executives confident about keeping individual agents in line, but unsure what happens when agents disagree, a gap that widens as firms chain agents together.

OpenAI buys Orchestra Bio to bolster its Rosalind life sciences AI: The four-year-old startup, which raised about US$20M from a16z and 8VC, builds agents for drug-discovery operations. Its team joins OpenAI’s Rosalind effort, already used by Moderna and Novo Nordisk.

THOUGHT LEADERSHIP

AI won’t fix messy processes; it just speeds up the mess: A Singapore AI marketing founder, drawing on a talk by Kaliber’s Robert Lai, argues most AI automates broken workflows instead of fixing them, so the disorder simply moves faster.

An AI trainer on turning a 40-hour week into 120 hours of output: The author says AI strips out the grind only when humans stay in the loop, and shares lessons on what to delegate, many of them learnt from AI itself.

A PR firm used Gemini to find which pitches actually pay off: After logging 300 hours on proposals in four months, the author compared five years of pitch proposals against client invoicing records to see which business was worth chasing.

AI took flawless workshop notes, then produced an unusable deck: A strategy workshop went well and an AI tool captured every point, but the resulting deck showed why synthesis still needs judgement, the author argues.

How one creator decides which part of an AI video workflow to fix: When LLM storyboards for a TikTok commerce creator produced loose shots, the author chose to diagnose the pipeline before rewriting scenes by hand.

A Vietnamese software firm audits its backlinks: only two count: The Ho Chi Minh City company assumed its directory listings helped search rankings, but an audit on 15 August found four domains linked, two counted for anything.

Google may already be showing its route to AGI: The author traces Demis Hassabis’s long-held view that AGI needs the right components, arguing Google’s recent moves reveal which pieces it now believes are in place.

The post Ecosystem Roundup: Taiwan blocks Grab’s US$600M foodpanda deal over Uber’s stake appeared first on e27.

Southeast Asia does not have a shortage of AI pilots. It has a shortage of AI pilots that ever leave the lab. A recent regional survey found that 57 per cent of SEA firms can’t scale their AI pilots, a stretch of corporate limbo now known in industry jargon as the “frozen middle”.

Red Hat, the IBM-owned open-source software company, thinks it can help thaw it. The firm on Thursday launched its APAC AI Centre of Excellence (CoE) in Singapore, backed by the Singapore Economic Development Board (EDB). Its first named collaborator is the Defence Science and Technology Agency (DSTA), the body that develops technology for the Singapore Armed Forces.

Also Read: Southeast Asian tech leaders are learning to trust AI agents, but not with production

The pitch is simple: help organisations move AI from experimentation to production using Red Hat’s open-source stack and engineering staff. The details are harder to pin down.

What the centre will actually do

Red Hat says the CoE’s engineers will work along two tracks. The first is building reusable software architectures, essentially templates for problems that keep recurring across industries, so that each client is not reinventing the plumbing. The second is contributing code “upstream”, meaning into the shared open-source projects that Red Hat’s commercial products are built on, rather than keeping fixes locked inside one customer’s deployment. The company names system deployment, runtime optimisation and platform management as priority areas.

This is familiar Red Hat territory. Over the past two years, the company has pushed hard on open-source inference tooling, betting that enterprises will want to run AI models across their own servers and multiple clouds rather than rent everything from a single hyperscaler.

“AI adoption is moving rapidly from experimentation towards execution, and organisations need open platforms and hands-on engineering expertise to make that transition responsibly,” said Daniel Aw, Vice President and General Manager of Red Hat Asia Pacific.

What Red Hat did not say is just as telling. The announcement carries no investment figure, no headcount, no timeline and no disclosure of what EDB’s “support” amounts to in grants or incentives. For a centre pitched as a regional hub, that leaves readers to take its scale on trust.

The defence angle

The DSTA tie-up is the most concrete part of the launch, and the most sensitive. The two organisations plan to jointly develop solutions in AI, cloud computing, application modernisation and edge computing, which refers to processing data close to where it is generated rather than in a distant data centre. They will also set up practices for DevSecOps, which builds security checks into software development from the start, and “Day 2 operations”, the unglamorous work of keeping systems running, patched and stable after launch.

The work will span on-premise and off-premise environments. In defence, that distinction matters: sensitive systems often cannot touch the public cloud at all, which is precisely where an open-source, run-it-anywhere vendor has an edge over hyperscalers.

“As threats become more complex, it is important that we continue to push the boundaries of technology and innovation to stay ahead of the curve,” said Gayle Chan, Deputy Chief Executive (Information) at DSTA.

Also Read: Singapore’s AI dividend will depend on what happens after the pilot phase

The language is still exploratory. DSTA is collaborating to “explore and co-develop” solutions it “can then apply” to defence uses. That is a long way from a deployed system. It also raises questions the announcement does not touch: what kinds of defence applications are in scope, and what oversight governs AI built for military use. A bank’s chatbot gets public scrutiny when it misbehaves. Defence AI rarely does, which is why clarity on guardrails matters more here, not less. The governance gap is not unique to defence; enterprises across the region still aren’t sure who referees their AI agents.

Selling talent as much as tech

Red Hat is also framing the CoE as a skills play, a theme that lands well with Singapore’s policymakers. The centre includes an innovation team of engineers, architects, project managers and forward-deployed engineers. The last category, popularised by Palantir, describes engineers who embed directly within client organisations rather than advising from afar, an admission that most AI projects fail on messy, company-specific problems rather than model quality.

The centre will also run training in software engineering, automation, cloud-native development and DevSecOps for customers and partners. That addresses a real bottleneck. Singapore firms are scrambling for staff who can build and manage autonomous systems, and agentic AI is now the city-state’s hottest job skill. Meanwhile, messy data keeps Singapore SMEs’ pilots stuck, a problem no platform alone can fix.

“The Centre bolsters Singapore’s position as a key AI hub creating opportunities for local talent to deepen their engineering capabilities,” said Yueh Chinn Seah, Vice President of Digital Industry at EDB.

A crowded rescue mission

Red Hat is far from alone in pitching itself as the cure for pilot fatigue. Earlier this month, Singtel’s RE:AI partnered the SIT-NVIDIA centre to tackle the enterprise AI deployment gap. Singapore-based OneByZero raised US$20 million in Series A funding for the same mission, while Thinking Machines joined the Temus group in a regional push for production-grade AI. Above them all sit AWS, Microsoft Azure and Google Cloud, each with deep pockets, regional data centres and its own enterprise AI programmes. Open-source rivals such as SUSE and Canonical compete for the same hybrid-cloud budgets.

Also Read: Everyone is talking about AI adoption. But are enterprises actually ready for it?

Red Hat’s differentiator is portability. Its argument is that enterprises and governments should not have to hand their AI future to a single cloud provider. That message resonates with the region’s growing anxiety that Southeast Asia cannot build sovereign AI on borrowed choices. It is worth noting, though, that “open” and “independent” are not the same thing: Red Hat is ultimately a US company owned by IBM.

The real test

Centres of excellence are easy to open and hard to judge. The proof will not be the ribbon-cutting but the case studies: how many regional clients move workloads into production, how much code the Singapore team ships upstream, and whether the DSTA work produces anything beyond press releases.

Until Red Hat puts numbers on the table, the CoE is a promising signal rather than a solved problem. For a region stuck in the frozen middle, even a signal counts, but enterprises would be wise to ask for receipts.

The post Red Hat opens Singapore AI centre to pull enterprises out of pilot purgatory appeared first on e27.

For a Singapore or Jakarta SaaS founder who has already sweated through GDPR paperwork to land a European client, Japan can look like a box already ticked. Strong privacy law, a mature regulator, a long history with data protection. Surely the European playbook will do?

It won’t. Japan’s Act on the Protection of Personal Information (APPI) shares a family resemblance with the EU’s General Data Protection Regulation (GDPR), but it handles consent, cross-border transfers and cloud hosting in ways that catch overseas startups out.

Also Read: Why Japan’s booming AI market is harder to crack than it looks

The stakes are rising. Japan is one of the world’s largest consumer markets, its corporates are hungry for partners, and its legal system is famously stable. That combination is drawing a steady stream of foreign founders, at the same time as Japanese companies are moving into Southeast Asia faster than many Western rivals. The traffic runs both ways, and personal data travels with it.

Who and what APPI covers

The law applies to “Personal Information Handling Business Operators”, a mouthful that covers almost any private company using a personal information database for business. Public-sector bodies sit under a separate regime.

APPI then sorts data into three tiers. Personal information is anything about a living person that can identify them, alone or when combined with other data. Personal data is that information once it sits in a structured database. Retained personal data is the subset a company has authority to act on when an individual makes a request. Data a company handles purely as a processor for someone else generally falls outside that last tier, which matters a great deal for SaaS vendors.

The obligations follow the data’s lifecycle. Companies cannot collect information through improper means, must state the purpose of use when collecting it, and cannot stray beyond that purpose. They must secure the data, supervise employees who touch it, and supervise any processors they hand it to. Sharing with third parties needs consent, with exceptions for outsourcing, business succession and joint use.

Individuals can ask a company to disclose, correct, add to or delete their retained data, to stop using it, or to stop sharing it with third parties.

Breaches carry a hard trigger. The company must notify both the Personal Information Protection Commission and the affected individuals when a leak involves sensitive information, could cause financial harm, looks malicious, or touches more than 1,000 people. For a SaaS platform with enterprise clients, crossing that last threshold is a bad afternoon, not a rare catastrophe.

Where GDPR muscle memory fails

The first trap is consent. Founders raised on GDPR expect to document a lawful basis for everything. APPI is in some ways lighter: ordinary personal information needs only a notice of purpose, not consent. Sensitive information is the exception, requiring prior consent before collection.

The second trap is cross-border transfers, where APPI leans much harder on consent than Europe does. Under GDPR, explicit consent is a narrow fallback behind adequacy decisions and standard contractual clauses. Under APPI, informed consent is a frontline option, but “informed” carries weight: the company must tell individuals the destination country, summarise that country’s privacy laws, and describe the recipient’s safeguards. The alternatives are sending data to a country Japan recognises as adequate, or a contract that binds the recipient to APPI-equivalent obligations.

Direct email marketing is the third. APPI expects prior consent, and a clause buried in the terms of service is discouraged, though an existing business relationship may help. Europe offers no single answer either, since the ePrivacy rules vary by member state.

Then there is infrastructure. Under APPI, moving data to a server abroad inside the same legal entity is not treated as a cross-border transfer, though the company must still review the foreign environment as part of its security duties. GDPR, by contrast, still demands safeguards for transfers within a corporate group.

Also Read: The compliance mosaic: Navigating SEA’s cybersecurity chaos

The biggest divergence is the “cloud exception”. If a cloud provider contractually agrees not to handle the personal data and enforces proper access controls, the cross-border and processor-supervision rules may not apply at all. GDPR has no such carve-out; every cloud provider needs an Article 28 processing agreement. For a lean startup, that exception could save months of legal back-and-forth, provided the contracts and access logs genuinely support it.

The SaaS reality check

Picture a foreign SaaS company selling to Japanese enterprises. The legal burden technically falls on the Japanese client, which must justify sending data abroad and must supervise its vendor by selecting it carefully, signing proper agreements and monitoring how data is processed.

In practice, that burden lands on the vendor’s desk. Japanese procurement teams will want terms of service that bind the provider to APPI-level obligations, so the client can rely on the contractual route rather than chasing consent from every employee or customer. A vendor that arrives with those terms ready shortens the sales cycle; one that arrives with a lightly edited GDPR template invites a long negotiation. Startups such as fileAI, which have expanded in Japan with corporate backers, offer a reminder that enterprise doors open through trust as much as product.

For Southeast Asian founders, this is familiar terrain. The region already juggles a patchwork of national data laws, and regulators such as Singapore’s are sharpening their focus on personal data in AI products. Japan simply adds another tile, with its own grain.

Trust is the real currency

APPI’s text is only half the story. Japanese companies and consumers have unusually high expectations around compliance, and those expectations have risen sharply. A regulatory slip can trigger reputational damage that outlasts any fine and quietly closes doors to the corporate partnerships that drew founders to Japan.

Also Read: Beyond market entry: Japan and Southeast Asia in a fracturing world

That is the uncomfortable takeaway for overseas startups: there is no shortcut by copy-paste. The founders who do well will treat APPI as a market-entry cost to budget for from day one, building a framework designed for Japan rather than retrofitting a European one. As privacy and sovereignty demands keep slowing enterprise AI deals across the region, getting this right early may be the most underrated go-to-market move a SaaS founder can make.

(This article was originally published on Black Box, a global media outlet that reports on the Japanese startup scene)

The post Japan’s privacy law isn’t GDPR, and that’s where foreign SaaS founders keep tripping appeared first on e27.

Anyone who has run security at a large Singapore organisation knows the routine. A vendor turns up with a slide deck, a slick demo and a quarterly target, and the talk is about the product long before anyone has asked what is actually keeping the security team up at night.

Raz Kotler has sat on the receiving end of that meeting more often than he would like. Now he is building a company that promises to do it the other way round.

Scouter, which launched in Singapore today, calls itself a “services-first” cybersecurity business. It pairs a short, curated list of security technologies with local engineers who help organisations choose, test, deploy and look after them.

Also Read: Singapore’s cybersecurity paradox: Leading in digital, lagging in defense

“Having spent years on the CISO side, including at PayPal, I know how frustrating it can be when someone tries to sell you a solution before understanding your problem,” Kotler said.

Most security chiefs will recognise the complaint. The harder question is whether Scouter can avoid becoming the thing it criticises.

From the buyer’s chair to the seller’s

Kotler is no newcomer to the region’s security circuit. Before Scouter, he had held several security leadership roles: Global Oversight CISO at PayPal, CISO at XM Cyber, head of security operations and engineering at an Israeli bank, and VP CISO at Deep Instinct.

Scouter is not a lone startup, either. It is the third Singapore business in a group that already includes Dart and Aleph, formerly known as Red Alpha. Together, the group claims nine years of work with Singapore enterprises, more than 30 government and enterprise customers, and more than 500 cybersecurity professionals placed through Aleph.

That history matters. Selling security to Singapore’s public sector and regulated industries is a relationship business, with long procurement cycles and very little patience for unfamiliar names. Scouter is new, but it is not starting from scratch.

What Scouter actually sells

The technology list is short: Wiz, Island, Oligo and Mate. The biggest name is Wiz, the cloud security platform that Google bought for US$32 billion in a deal that closed on March 11, 2026. Island makes an enterprise browser that puts security controls inside the browser itself. Oligo focuses on protecting applications while they run.

Around these products sit four service lines: Managed Detection & Response (MDR), which means outsourced round-the-clock monitoring and incident handling; Forward-Deployed Cyber Engineer; Architecture, Validation & Implementation; and Enablement & Lifecycle Support.

The forward-deployed engineer is the most interesting of the four. Palantir made the model famous: an engineer works inside the customer’s environment instead of handing over a product and leaving. In security, where tools often sit half-configured on a shelf, that could make a real difference.

Also Read: The demand for SMB cybersecurity is inevitable, the supply was never built correctly

The core pitch is validation before implementation. Scouter says it will show that a tool works in a client’s own environment before the client commits to it, and then support it after deployment.

The timing helps. Attackers are getting cheaper and faster, with AI agents now able to break into companies for pocket change. Regulators are also paying closer attention: the Monetary Authority of Singapore and the Association of Banks in Singapore recently formed a taskforce on AI-driven cyber risk. Buyers have less money and less time for tools that disappoint.

The independence question

There is a tension here that the launch announcement does not address. Scouter presents itself as starting from the customer’s problem. But it also has a curated portfolio of four vendors, and that is still a catalogue. An adviser that also earns money reselling the products it recommends has a conflict of interest, however good its intentions.

The release also leaves out several things a buyer would want to know: how Scouter makes money, what it charges, who its first customers are, how many people it employs today, and whether it has outside funding. It is fair to ask what Scouter will tell a client whose best answer is not on its list, or whose best answer is to buy nothing new at all.

A crowded middle market

Scouter is entering a busy part of Singapore’s security market. At the top end are the heavyweights: Ensign InfoSecurity, ST Engineering’s cyber arm and Singtel-owned NCS. These firms bundle advisory, managed services and integration for government and large enterprises. Global managed security providers and value-added resellers compete for the same budgets. Regional distributors supply much of the hardware and software. The vendors themselves increasingly sell professional services directly, and Wiz now sits inside a Google Cloud that also owns incident-response specialist Mandiant.

Scouter’s advantage is meant to be focus: a buyer’s-eye view from a former CISO, a deliberately small portfolio, and engineers embedded with the client rather than a sales team. Whether that is a lasting position or just a niche will depend on how many of the group’s 30-plus existing customers decide to sign up.

The talent bet

The part of the launch with the most potential is also the least detailed. Scouter says it is building Singapore’s AI security capability by training its current staff, which is already underway, and by hiring more engineers. It has not given any numbers for either.

The need is real. In 2025, Red Alpha itself pointed to an estimated shortfall of nearly 4,000 cybersecurity professionals in Singapore. AI security is a newer field still, and few people in the region have hands-on experience securing models, agents and the data pipelines that feed them. Singapore has written strong rules, but reports suggest it struggles to enforce them, and that the skills gap remains the bottleneck.

Aleph’s background in training people for security jobs gives Scouter some credibility on this front. A promise to build local skills is easy to make, though. The real test will be how many engineers Scouter hires, how they are trained, and whether clients’ own teams end up able to run the technology without Scouter.

Also Read: AI is already inside the enterprise. Has security kept up in Asia?

Kotler has identified a frustration that nearly every CISO in the region shares. If Scouter’s engineers are willing to tell a client “you don’t need this,” it may stand out in a market built on selling. If they are not, it will be one more reseller.

The post Scouter enters Singapore’s crowded cyber market with a “listen first” promise appeared first on e27.

Ask a room of enterprise executives whether they can keep their AI agents in line and most will nod. Ask them what happens when two of those agents disagree, and the room goes quiet.

That, in essence, is the finding of a new report from OutSystems, the Lisbon-born low-code company that now markets itself as an “Agentic Systems Platform”. Produced with trade publication CIO Dive and consultancy KPMG, Engineering Business Outcomes: The Shift to Agentic Systems surveyed more than 850 IT decision-makers. Some 82 per cent said their organisations can keep pace with governing, integrating and monitoring AI agents as they move into production. Only 26 per cent said they were well prepared to resolve conflicts across agents, systems and automated workflows.

Also Read: The hidden economics of autonomous AI agents

That 56-percentage-point gap is the story. Confidence is cheap. Conflict resolution is not.

What an agent fight looks like

An AI agent, unlike a chatbot, does not just answer. It acts: it updates a record, triggers a refund, books a shipment. Put several of them inside one company and they start treading on each other’s toes. A procurement agent trims inventory while a sales agent promises next-day delivery. A risk agent freezes an account that a customer-service agent has just reactivated. Each is following its own rules. Together, they create a mess, and someone has to decide which action stands and how to unwind the other.

The report suggests most enterprises are not wired to make that call quickly. While 83 per cent rely on existing IT, security or compliance processes, specialised reviews or centralised policies to oversee agents that can act autonomously, just five per cent have automated controls, audit trails and formal human review points in place. Oversight, in other words, is still largely a person reading logs after the damage is done.

Southeast Asia has seen this film before

None of this will surprise anyone tracking enterprise AI in the region. In June, research by Sinch found that Singapore led Asia Pacific in deploying AI agents and also in pulling them back after going live, with 82 per cent of the city-state’s enterprises reporting a rollback or shutdown. In August, a Sumsub and Singapore Fintech Association study found that 94 per cent of Singapore businesses were using or piloting multi-step AI systems, yet only 29 per cent could produce an audit trail for AI-driven decisions.

The engineering floor is warier than the boardroom. The Agoda AI Developer Report 2026, which polled more than 800 developers across Southeast Asia and India, found that 79 per cent require human approval before AI-generated work goes live. Agents are trusted, but not with production. Thailand’s SCB 10X even caught agents confidently declaring tasks complete when the underlying code had failed its specifications.

Regulators have taken note. Last week, Singapore’s Infocomm Media Development Authority (IMDA) unveiled a starter kit for testing agentic systems and an accreditation scheme for third-party AI testers, effectively building an industry whose job is to check the agents. One of the kit’s metrics, handoff accuracy, measures whether an agent passed a task to the right agent or human. That is precisely the seam where the OutSystems respondents admit they are weakest.

The data problem underneath

Agents collide most often when they are working from different versions of the truth. More than a third of respondents (35 per cent) cited data consistency and context sharing as a major orchestration hurdle, which is a polite way of saying agents are too often running on generic model knowledge rather than governed company data and live application states.

Also Read: Agentic AI is Singapore’s hottest job skill, and that should worry the people it replaces

Singapore has its own version of this headache. A Confluent survey in July found that 78 per cent of the city-state’s IT leaders flagged a lack of real-time data infrastructure, leaving agentic ambitions stuck on legacy systems and data quality gaps.

Buyers also want an exit door. Some 74 per cent prioritise integration with existing applications, 63 per cent want agent orchestration across systems, and 62 per cent value the ability to switch models, vendors and platforms over time. After a decade of cloud lock-in, enterprises are in no mood to marry a single AI supplier.

Read the fine print

A dose of scepticism is warranted. This is vendor-commissioned research, and its central prescription, that governance must be built into the platform and enforced while agents run rather than reviewed afterwards, happens to describe what OutSystems sells.

Tiago Azevedo, the company’s CIO, makes the pitch plainly: “Safety comes from building control into the platform, not reviewing it afterward.” KPMG Managing Director Michael Harper argues that logging and procedural governance were designed for static software, and that enterprises now need policy-as-code and runtime guardrails. Both points are reasonable. Both also lead neatly to a purchase order.

The outcome data is more revealing than the quotes. Error reduction (64 per cent) is the leading measure of impact, followed by compliance and risk reduction, productivity and cost. Yet only 12 per cent say agentic AI has sped up the launch of new products, services or customer experiences. Agents are tidying up existing operations, not inventing new businesses. Meanwhile, 81 per cent rate testing and validating agent behaviour as very or extremely important when choosing a platform, which tells you how little they trust what is already running.

Context matters, too. Last year’s edition of the same study, based on 550 software executives, found 93 per cent of organisations building or planning custom agents. The ambition has arrived. The plumbing has not.

A crowded referee’s chair

OutSystems is hardly alone in offering to referee. Salesforce is pushing Agentforce, Microsoft has Copilot Studio, ServiceNow and UiPath are pitching orchestration layers, and low-code rivals such as Siemens-owned Mendix and Appian are making near-identical governance promises. Each claims its platform is the safest place for agents to work together. The 62 per cent of buyers who want the freedom to switch suggest few are ready to hand the whistle to any single vendor.

In Southeast Asia, banks are drawing the line in their own way, as UOB’s research shows: AI agents can move money, as long as someone is accountable. Even model makers are wrestling with the problem; OpenAI shelved its GPT-6.1 Astra model this month amid concerns over how honestly the agent reported its own work.

Also Read: Moving past the chatbox: The hidden risks of agentic AI and MCP in enterprise infrastructure

The harder question is who referees the agents at companies that cannot afford a KPMG engagement. A mid-sized retailer in Jakarta or a lender in Manila deploying off-the-shelf agents will rely on whatever controls its vendors ship, plus free templates such as IMDA’s. As AI governance moves from promises to proof, that is where the next failures, and the next regulations, are likely to emerge.

Eighty-two per cent of executives believe they can govern their agents. The other number, 26, is probably closer to the truth.

The post Who referees the AI agents? Enterprises aren’t sure, new report shows appeared first on e27.

Vingroup is opening a third Asian manufacturing front for its electric vehicle (EV) ambitions. The Vietnamese conglomerate, best known for EV maker VinFast, plans to invest US$500 million in a flagship EV plant in the Philippines, the country’s Department of Trade and Industry (DTI) said on Wednesday.

The factory is designed to produce 200,000 vehicles a year at the outset, for both domestic buyers and export markets.

Construction is slated to begin in the first quarter of 2027 and take 12 to 18 months, which puts the earliest realistic start of production somewhere in 2028. Alongside four-wheelers, Vingroup also plans to bring electric scooters to the Philippine market.

Also Read: EVs gain ground in the Philippines as fuel costs squeeze drivers

The DTI pitched the project as more than a car factory, describing it as a possible base for Vingroup’s expansion into multiple industries in the country. Trade Secretary Cristina Roque said the investment reflects the Philippines’ growing status as a Southeast Asian hub for advanced manufacturing and green technology, and promised “end-to-end support” for investors.

The maths behind 200,000 units

The headline capacity deserves a second look. BMI Country Risk and Industry Research expects the Philippines to sell 423,750 vehicles in total in 2026, of which only 32,776 will be EVs, a figure that includes plug-in hybrids. On those numbers, the plant could build about six times the country’s entire projected EV market for the year, and close to half of all new cars Filipinos are expected to buy. Even BMI’s more bullish 2030 forecast of 91,730 EV sales leaves the factory with plenty of idle capacity.

The export clause, in other words, is the business case. ASEAN’s free trade framework allows vehicles that meet regional content rules to move across the bloc without tariffs, and the Philippines sits on busy shipping routes to the rest of Southeast Asia and beyond.

There is a policy wrinkle, too. Battery EVs imported into the Philippines are currently exempt from import tariffs and excise taxes until 2028, roughly when the plant is due to come online. That means a locally built VinFast would get little tariff protection against imported Chinese models in its early years. Its advantage must come from elsewhere: tax breaks under the CREATE MORE Act, which extends incentives such as enhanced deductions and lower corporate income tax rates for registered projects, and faster permitting through the “Green Lanes” set up under Executive Order No. 18 for strategic investments. If the import exemptions lapse after 2028, a local factory becomes considerably more valuable. The timing may be less coincidental than it looks.

A familiar playbook, now in its third country

The Philippine plant follows a template VinFast has already rolled out twice. In Indonesia, it opened a US$300 million factory in Subang, West Java, about 120 km south-east of Jakarta, in December 2025, and said it would spend US$1 billion to expand capacity to 350,000 vehicles a year. In India, where its plant is in Tamil Nadu, it announced plans that same month to invest US$500 million to triple output.

Manila’s courtship has been a long one. In January 2024, President Ferdinand Marcos Jr met Vingroup chairman Pham Nhat Vuong in Hanoi, where the conversation revolved around dealerships, battery production and nickel processing. Two and a half years later, it has become a factory.

Also Read: BYD co-founder-backed OMOWAY bags funding to join Southeast Asia’s electric motorcycle race

The scooter plan may prove the more interesting bet. Two-wheelers and tricycles are the workhorses of Philippine streets, and Marcos has repeatedly tied EV policy to replacing ageing jeepneys and tricycles. VinFast has form here: its ride-hailing affiliate Green SM has taken e-scooters into Jakarta’s ojek economy, and the company has experimented with rentals to get around Southeast Asia’s affordability problem.

Rivals already in the garage

VinFast will not have the road to itself. Chinese brands, including BYD, MG, GAC Aion and Chery, are already pushing EV prices down in the Philippines with affordable SUVs and crossovers. BYD has also built manufacturing capacity in Thailand, which has positioned itself as a regional EV production hub, while Indonesia has used its nickel reserves to lure battery and EV makers. Japanese incumbents such as Toyota and Mitsubishi, which assemble vehicles locally, still dominate the Philippine market and are leaning on hybrids, a format many buyers prefer while charging remains scarce.

At home, VinFast is a heavyweight, accounting for 42 per cent of Vietnam’s domestic car sales in August 2026, helped by its own charging network and taxi fleets. Abroad, it is a challenger with a modest brand, competing in a regional EV race that has become crowded quickly. Even its home turf is being contested, with Tesla now incorporated in Vietnam.

Can the balance sheet carry it?

The bigger question is money. VinFast reported unaudited second-quarter 2026 revenue of VND30,340.9 billion (about US$1.2 billion), up 85 per cent year-on-year. That is impressive growth. But its loss from continuing operations widened 128 per cent to VND31,189.2 billion (about US$1.2 billion), which means it lost slightly more than it earned.

The total net loss of VND1,414.9 billion (about US$54 million) looks far healthier, but only because of a gain from discontinued operations. That is an accounting effect, not evidence that the core business is turning around.

Also Read: Tesla establishes Vietnam subsidiary as EV rivalry with VinFast looms

VinFast has long relied on financial backing from Vingroup and its founder to fund its expansion, and the Philippine project is being announced under the parent’s banner. That may be deliberate. The DTI’s talk of a multi-industry base suggests the conglomerate, not just the carmaker, is putting its name to it.

For the Philippines, which is still building out a network of only about 1,600 charging points, the plant is a welcome signal that it can compete with Thailand and Indonesia for EV manufacturing. Whether that signal turns into reality depends less on ribbon-cuttings in 2028 and more on whether the ships leaving Philippine ports are full.

The post A 200,000-car plant for a 33,000-EV market: Inside VinFast’s Philippine gamble appeared first on e27.

From October 27 to 29, 33 Taiwanese startups will exhibit at the Taiwan Pavilion at SWITCH 2026 in Singapore (Booth Q01, Sands Expo & Convention Centre), as part of Taiwan Tech Week in Singapore. The delegation is led by Startup Island TAIWAN, Taiwan’s national startup brand backed by the National Development Council (NDC), in partnership with Startup Terrace, Hsinchu Science Park Bureau, the Taiwan Stock Exchange (TWSE) and Taiwan Digi Gold Card.

Southeast Asia’s innovation economy is moving from experimentation to execution. Enterprises across the region are looking for technology that can be piloted quickly, integrated into existing operations, and scaled across borders.

That is the opportunity the delegation is bringing to Singapore. It spans four sectors: manufacturing and semiconductors, AI and software, healthcare and biotechnology, and energy and sustainability. Many of the startups are already running pilots or serving paying customers, and they are coming to SWITCH to meet enterprise partners, distributors, and investors from across the region.

A look at who’s exhibiting

Manufacturing, hardware and semiconductors

AiXaS (active interposerX as service)

Taiwan-based semiconductor startup AiXaS (active interposerX as service) supports physical-AI design. At SWITCH 2026, AiXaS is looking for system migration clients and chiplet supply-chain counterparts.

Alpenglow Tek

Taiwan-based AI chip startup Alpenglow Tek provides ultra-low-power AI chips that help device manufacturers enable always-on AI sensing and real-time interaction with 100x lower power consumption. In 2026, Alpenglow Tek won the Gold Award in the IC Design category for Startups and SMEs at Taiwan’s 2026 Smart Innovation Awards, and has 3 paying customers and 10 potential customers in its sales pipeline. At SWITCH 2026, Alpenglow Tek is looking to meet wearable device and robotics manufacturers looking to integrate on-device AI into their products.

Hyarks

Taiwan-based marine technology startup Hyarks develops modular autonomous Unmanned Surface Vehicles (USVs) and integrated platforms to drive safe, sustainable, and efficient maritime operations for commercial fleets and offshore operators. In 2026, Hyarks secured deployment agreements for its TF-100 USV platform with 5 industry leaders. At SWITCH 2026, Hyarks is looking to meet investors.

Leafy Lab

Deterministic AI for every chip that powers the physical world. Taiwan-based semiconductor AI startup Leafy Lab serves 6 enterprise clients in 2 countries, with technology more than 10x faster than current manual processes. At SWITCH 2026, Leafy Lab is looking for customers.

Memorence AI

Taiwan-based industrial AI startup Memorence AI provides reliable instant-learning AI that helps high-mix manufacturers improve vision-language guidance, inspection, and traceability across real-world production environments. Memorence AI was named a CES Innovation Awards 2026 Honoree in Enterprise Tech and won the Embedded Vision category at the embedded award 2026. At SWITCH 2026, Memorence AI is looking for customers among manufacturers in electronics, semiconductors, medical devices, aerospace, and other high-mix, high-value industries.

NanoKnife Semiconductor Technologies Co., Ltd.

Taiwan-based semiconductor startup NanoKnife Semiconductor focuses on improving Through-Silicon Via, or TSV, fabrication for advanced semiconductor packaging. At SWITCH 2026, NanoKnife Semiconductor is looking for technology and R&D partners.

Nudesign

Taiwan-based aviation technology startup Nudesign develops innovative 4-stroke aviation powertrains that deliver superior fuel efficiency, reliability, and payload performance for drones and light aircraft. Nudesign has grown from one customer to 2 customers, with about a 10 per cent revenue increase over last year. At SWITCH 2026, Nudesign is looking to run PoCs with drone manufacturers, aerospace companies and urban air mobility innovators.

SiCEV

Taiwan-based power semiconductor startup SiCEV provides customised power device design services and works closely with customers to enhance product performance and accelerate time-to-market. SiCEV has raised NT$200 million and has been granted 12 Taiwan invention patents. At SWITCH 2026, SiCEV is looking to meet potential customers.

Tera Electronics

Taiwan-based optical interconnect startup Tera Electronics provides MicroLED optical interconnect solutions that help AI data centre and HPC system developers increase data transmission bandwidth and reduce power consumption. In 2026, Tera Electronics raised over NT$300 million to advance MicroLED optical interconnect development and commercialisation, and joined the NVIDIA Inception program. At SWITCH 2026, Tera is seeking Southeast Asian partners to advance joint development and commercialisation for AI data centers and HPC systems.

WEAVEN

Taiwan-based AIoT startup WEAVEN provides AIoT watch boxes generating Digital Health Resumes to help luxury watch collectors optimise servicing, helping reduce blind trading risks and maximise resale value. WEAVEN has accumulated over 35,000 unique movement health data points and co-hosted exclusive events with BVLGARI. At SWITCH 2026, WEAVEN is looking to meet channel partners and potential customers.

Also Read: Taiwan bets on Gen Z founders to move beyond its chip-supplier image

AI and software solutions

Aira

Taiwan-based AI startup AIRA provides CPU-optimised AI video analytics solutions that help enterprises, system integrators, and security operators enhance safety, operational efficiency, and intelligence using existing camera infrastructure. AIRA’s smart factory solution reduced attendance processing time from 16 hours to 1 hour, while increasing equipment utilisation by 50 per cent and production capacity by 53.26 per cent. At SWITCH 2026, AIRA is looking for local system integrators to enable on-site deployment and customer support.

Asiabots Limited

Taiwan-based AI startup Asiabots provides AI Ambassador, AI Voicebot and AI Chatbot that help enterprises deliver multilingual customer service across physical locations, phone calls and digital channels. In 2025, Asiabots received the 24th Startup Award from Taiwan’s Ministry of Economic Affairs (MOEA). At SWITCH 2026, Asiabots aims to explore partnerships for local pilots and regional deployment with enterprise decision-makers in retail, hospitality, transportation and public services.

Cellbedell

Taiwan-based smart building startup Godspeed provides Cellbedell, a Smart Building OS combining virtual keys, edge verification and Nexus infrastructure to help commercial spaces automate access and unattended operations. By August 2026, Cellbedell had more than 250 partners and business clients. At SWITCH 2026, Godspeed is seeking Singapore enterprise partners and pilot sites to expand across offices, co-working spaces and hospitality in Southeast Asia.

eCloudEdge / NeoEdge

Taiwan-based industrial AI startup eCloudEdge provides NeoEdge, a data governance and industrial AI orchestration platform that helps manufacturers connect and govern OT/IT data, building domain-specific AI agents to optimise processes and reduce costs. eCloudEdge was included in STL Partners’ “50 edge computing companies to watch in 2026” list. At SWITCH 2026, eCloudEdge is looking for Singapore-based OT system integrators and industrial automation solution providers who can co-deliver NeoEdge to local manufacturers.

futurenest

Taiwan-based enterprise AI startup futurenest builds the foundational infrastructure for enterprise intelligence: Xerno governs what AI knows and Xparse governs how AI acts, so regulated enterprises can deploy intelligence at scale without surrendering control over it. futurenest has 12 signed enterprise customers, all contracted and in active deployment across three regulated domains: telecom, national defense and national transport. At SWITCH 2026, futurenest is looking for enterprise partners.

LMT

Taiwan-based fintech startup Light Momentum Technology provides ultra-low-latency FPGA trading infrastructure and AI financial automation that help financial institutions accelerate execution, optimise operations, and improve system efficiency. LMT has worked with leading Taiwanese securities firms on low-latency trading systems, helping optimise execution performance and system stability. At SWITCH 2026, LMT is seeking strategic partners for overseas deployment, co-development and market entry.

MantaGO

Taiwan-based AI startup Eslitec builds AI driven CRM ecosystem tailored for SMEs, making customer engagement simpler and more strategic. Eslitec has 2,280 clients worldwide and has partnered with 25 distributors. At SWITCH 2026, MantaGO is looking for distributors and channel partners.

Next Integral

Taiwan-based edge AI startup Next Integral develops NEXUO, a modular Edge AI platform transforming existing cameras into real-time safety monitoring and operational intelligence for factories, smart facilities, logistics, and care environments. Next Integral was selected for CES USA 2026, VivaTech France 2026, IFA Germany 2026 and Fly Asia Korea 2026. At SWITCH 2026, Next Integral is seeking Singapore partners to validate NEXUO in local environments through 60–90 day PoCs.

Optiqb

Optiqb AI Technology is a Taiwan-based AI startup specialising in 3D vision, bringing a new era of 3D vision in the mobile age through AI computer vision and optical technology. Optiqb won Best in AI and Robotics at the Global Deep Tech Battle at VivaTech 2025. At SWITCH 2026, Optiqb is looking for enterprise partners and content providers.

Speculum AI

Taiwan-based AI startup Speculum AI provides an AI runtime layer that helps organisations building and operating LLMs, agents and physical AI achieve greater stability, control and efficiency across cloud, on-prem and edge environments. Within six months of entering the market, Speculum AI secured an enterprise PoC with a major Japanese system integrator and established local operations in Japan. As it plans its expansion into Singapore, Speculum AI aims to grow its customer and channel partner network through SWITCH 2026.

KABOB Cloud- StreamActivate

Taiwan-based data analytics startup StreamActivate provides a real-time data management and analytics platform that helps enterprises turn live data into actionable insights, respond faster, and build an AI-ready data foundation. In 2026, technical validation with Coretronic demonstrated a 53 per cent reduction in combined inspection, measurement, and reporting time, from 1,616 to 755 minutes. At SWITCH 2026, StreamActivate aims to connect with regional system integrators and channel partners.

Tenfold AI

Taiwan-based legal AI startup Tenfold AI provides a civil-law legal AI platform LexGents that helps APAC lawyers research, draft, and review legal work with trusted local-law grounding. Tenfold AI serves 20+ corporates and law firms with 300+ users across 4 countries, and has raised US$300K. At SWITCH 2026, Tenfold AI is looking to secure PoCs with government, in-house legal teams and small law firms.

Also Read: Entering Taiwan: A practical resource map for foreign founders

Healthcare and biotechnology

Coherence Bio

Taiwan-based biotech startup Coherence builds Keystone, a benchtop continuous lab automation system that helps organoid producers and drug discovery teams keep multi-day cell and organoid experiments on one controlled timeline and record. In 2026, Coherence Bio signed a pilot agreement and letter of intent with a Korean organoid manufacturer and won first prize in the medical-device category for startup companies at the 2026 NBRP Demo & Pitch Day. At SWITCH 2026, Coherence Bio is looking to secure PoCs with organoid and cell-model producers, research institutes and AI drug-discovery teams, and to meet life-science and deep-tech investors.

EpiSante Biomed

EpiSante Biomedical is a Taiwan-based biotechnology company developing precision medicine technology based on DNA methylation platforms, including a liver cancer methylation test kit for early detection and disease monitoring. At SWITCH 2026, EpiSante Biomed is looking to meet investors.

Fluidiconic Biotechnology

Taiwan-based biotech startup Fluidiconic provides MedSelect®, a patient-derived tumor platform that helps drug developers evaluate treatment responses in human-relevant models and select promising cancer therapies for further development. Fluidiconic has 19 paying customers, including pharmaceutical companies and research centers, and its drug-response results showed 91.6 per cent agreement with animal-study outcomes. At SWITCH 2026, Fluidiconic is looking to meet investors and is seeking Singapore and Southeast Asian partners to bring patient-derived drug-response evidence into cancer research.

ible

Taiwan-based health tech startup ible Technology provides wearable air purifiers and air purification solutions that help individuals reduce exposure to airborne pollutants/ viruses and enjoy cleaner air wherever they go. ible has received two CES Innovation Awards recognitions and one German Innovation Award, and has established sales presence in Japan and Switzerland. At SWITCH 2026, ible is looking for distribution and retail partners to expand into Southeast Asia.

LongGood

Taiwan-based digital health startup LongGood provides digital rehabilitation that helps therapists, physicians, and patients keep on track with their rehabilitation journey. LongGood has customers across Taiwan, Japan, Singapore, Thailand, and Hong Kong. At SWITCH 2026, LongGood is looking for distributors or channel partners.

ST Medical

Taiwan-based medical AI startup ST Medical Technology develops AI-powered clinical research and medical data platforms that help hospitals and biotech teams accelerate analysis, collaboration, and healthcare innovation. ST Medical serves 20+ enterprise, healthcare, and research clients across 4 countries. At SWITCH 2026, ST Medical is looking for distributors or channel partners as it builds new partnerships in Europe and Southeast Asia.

Also Read: AI could turn concierge medicine into the future interface to healthcare

Energy and sustainability

Ancillary Power Corp. Ltd.

Taiwan-based energy tech startup Ancillary Power provides an AI-powered Virtual Power Plant platform and green energy trading solutions that help industrial clients and RE100 enterprises to optimise electricity costs and achieve net-zero targets. Ancillary Power has aggregated over 270 MW of distributed energy resources, and its revenue grew from NT$133M in 2024 to NT$334M in 2025. At SWITCH 2026, Ancillary Power is seeking enterprise partners among telecoms, hotel franchises, energy companies and AI data centres.

Apollo Power

Taiwan-based energy tech startup Apollo Power delivers AI-driven energy management and renewable energy services, with an AI dispatch system that helps energy-intensive industries cut costs and strengthen supply stability. Apollo Power registered on Taiwan’s Emerging Stock Board on April 28, 2026, and it manages over 100 GWh of cumulative green electricity procurement and dispatch. At SWITCH 2026, Apollo Power is looking to secure PoCs with Southeast Asian enterprises, commercial property developers, and regional sustainability leaders.

AquaOptima

Taiwan-based water tech startup AquaOptima provides an AI Operator solution for water utilities that helps water treatment plants, pumping stations, and distribution networks run pumps autonomously — reducing energy consumption and minimising water loss. AquaOptima has 2 active deployments with Taiwan Water Corporation, and has achieved 15–20 per cent energy savings at the Yilan water treatment plant, with third-party measurement and verification by Carbon Trust ongoing. AquaOptima is actively closing its seed round, and investor introductions and meetings are welcome at SWITCH 2026.

CarbonClean Energy

Taiwan-based climate tech startup CarbonClean Energy provides modular CO₂ electrolysis systems that help industrial manufacturers convert captured CO₂ into syngas and green chemicals, enabling practical industrial decarbonisation and circular carbon utilisation. In 2026, CarbonClean Energy completed a 3-week continuous field demonstration using real industrial flue gas and established industrial collaborations with Tung Ho Steel and Chang Chun Group. At SWITCH 2026, CarbonClean Energy is looking to secure a PoC as it explores partnerships in Southeast Asia.

3Egreen (Extreme Energy Efficiency)

Taiwan-based energy tech startup 3Egreen provides Edge-AI energy monitoring IIoTand predictive maintenance solutions that help high-tech and industrial manufacturing plants optimise power efficiency and prevent equipment downtime. 3Egreen has deployed its Edge-AI energy monitoring and asset diagnostics solutions across leading high-tech and semiconductor manufacturing ecosystems, including Tier-1 global memory manufacturers and key regional system integrators. At SWITCH 2026, 3Egreen is looking to meet power and utility companies and ESG consultants.

Meet them at SWITCH 2026

The Taiwan Pavilion is open from October 27 to 29, 09:00–18:00, at Sands Expo & Convention Centre, Level 1, Booth Q01. Visitors can meet the founders, explore partnership opportunities, and view the startups’ one-pagers here.

—

Want updates like this delivered directly? Join our WhatsApp channel and stay in the loop.

The post Taiwan Pavilion at SWITCH 2026: Meet the 33 Taiwanese startups heading to Singapore appeared first on e27.

Indonesia’s D2C and inventory-heavy startup sector looks different in 2026 than it did before 2022. The cohort of platforms that scaled between 2018 and 2021 (Fabelio in furniture, several in F&B, fashion, and household goods) encountered a category of operational risk that the original venture playbook tended to underestimate. Some platforms have stabilised. Several have restructured significantly. Fabelio, which entered PKPU proceedings in 2022 and was subsequently declared insolvent, is one of the more publicly documented examples.

After 15 years inside Indonesian risk functions, I have watched this sector with more interest than the financial services focus would suggest. The operational risk patterns underneath the D2C failures of this period are not unique to those sectors. They recur in any business where the gap between cash going out and cash coming back is large, and the venture funding model assumes that gap can be financed indefinitely.

What the D2C generation aimed to do

The Indonesian D2C cohort that emerged after 2018 had a clean thesis. Indonesia’s rising middle class, expanding e-commerce infrastructure, and digital payment penetration created a window for direct-to-consumer brands to skip traditional retail intermediation. The implicit assumption was that unit economics would improve as scale increased: customer acquisition costs would decline, operational costs would compress, working capital intensity would moderate.

What turned out to be harder

Three structural factors made the path harder than the original models priced in.

The cash conversion cycle. D2C businesses convert capital into inventory, inventory into delivered orders, and delivered orders into recognised revenue across a cycle structurally longer than software businesses face. A platform that grows 100 per cent in revenue often grows working capital requirements faster than that, especially when product categories carry inventory and returns risk.

Also Read: The agritech credit paradox: Lessons from TaniHub and Indonesia’s first agritech generation

Inventory as a financial product. The inventory on a D2C platform’s balance sheet is not just stock. It is essentially a leveraged position on consumer demand at the price and SKU mix chosen six to 12 months earlier. When demand shifts, the inventory becomes more expensive to liquidate than to acquire. Platforms that priced inventory as a working capital line item rather than as a market exposure mispriced one of their largest risks.

Operational fragility at scale. Operational systems that work for a hundred orders a day break down at a thousand orders a day in ways not always visible until they break. The platforms that scaled fastest sometimes did so on top of operational systems that were not actually ready for the volume, with quality and customer experience consequences that compounded into churn and brand damage.

Lessons learned

Six principles emerge from the Indonesian D2C cycle.

Working capital is the business model. For inventory-heavy D2C, working capital management is not back-office. It is the business. Platforms that treated cash conversion cycle, payment terms, and inventory turn as primary metrics outperformed those that treated them as supporting metrics.

Growth without unit economics is a trap. The venture-funded D2C generation often raised on the assumption that scale would deliver unit economics later. Several categories never reached the inflection point because the cost structure was wrong from the beginning. The unit economics question should be answered before scale, not in spite of scale.

Inventory needs a hedging instinct. Holding inventory is taking a position on future demand. The platforms that built inventory planning, demand forecasting, and discount management as core disciplines absorbed inventory risk better than those that treated inventory as procurement.

Also Read: Same failure, better clothes: Matchmade.io on selling Indonesian finance tech in Singapore

Operational systems must scale before volume scales. The institutions that survived the volume ramp had invested in operational infrastructure before the volume arrived. The ones that scaled volume first and operations second produced the customer experience failures that drove churn.

Founder accountability matters more in physical businesses. In software, a founder’s bad month produces a feature delay. In inventory-heavy D2C, a founder’s bad month can produce a cash flow event that becomes an insolvency event inside a single quarter. The governance scaffolding around the founder matters more when the business has physical exposure than when it does not.

PKPU is a recovery instrument, not just a failure marker. The Indonesian PKPU framework is designed to give viable businesses a structured path to restructure. The companies that struggled most were the ones where governance and disclosure had already eroded before PKPU was filed, not the process itself.

The macro stakes

Indonesia’s D2C category will not disappear. The structural opportunity (large rising middle class, growing e-commerce infrastructure, persistent inefficiency in legacy distribution) remains real. The next generation will be more capital-disciplined, more operationally rigorous, and more humble about the difference between software-style and physical-goods-style business models.

For founders and investors looking at D2C and inventory-heavy categories in 2026, the question is no longer whether the opportunity exists. It is whether the platform has built the working capital discipline, operational infrastructure, and governance scaffolding required to convert the opportunity into a durable business, rather than into another addition to the post-mortem literature.

—

The post The working capital trap: Lessons from Indonesia’s first D2C generation appeared first on e27.