AGIBOT, a Chinese robotics company specialising in embodied artificial intelligence, has launched operations in Malaysia, marking the beginning of its Asia-Pacific expansion strategy.

The company held a launch event on Tuesday at i-City in Selangor, attended by Malaysian government officials and industry partners, coinciding with the opening of an AI and Robotics Experience Centre developed with local property developer I-Berhad.

Also Read: SGInnovate backs Botsync in extended Series A amid AMR market surge

The timing reflects broader shifts in Southeast Asia’s manufacturing landscape. Rising labour costs, supply chain reshoring, and government-backed automation initiatives are driving demand for robotics solutions across the region.

Malaysia’s established electronics manufacturing base, relatively business-friendly regulatory environment, and central geographic position make it a logical entry point for regional expansion.

Market position and scale

AGIBOT, founded in 2023, has achieved rapid scaling. The company claims to have shipped over 5,100 humanoid units in 2025 and was ranked No. 1 globally by market research firm Omdia for humanoid robot shipments and market share, capturing 39 per cent of the global market. The company’s 5,168th mass-produced unit demonstrates industrial-scale manufacturing capability—a significant milestone for a robotics sector historically characterised by limited production volumes.

“AGIBOT made significant strides to improve the mass production and the practical use of embodied robotics last year,” said Deng Taihua, Founder, Chairman, and CEO. “This milestone puts AGIBOT in a strong position as we start 2026.”

AGIBOT’s approach differs from traditional industrial robots designed for single, repetitive tasks. The company focuses on embodied AI systems that learn and adapt in real-world environments through reinforcement learning, enabling robots to be trained and deployed directly in operational settings rather than relying solely on pre-programmed sequences.

Malaysia partnership and testbed model

The AI World Experience Centre represents a departure from conventional robotics commercialisation. Rather than focusing exclusively on factory automation, AGIBOT and I-Berhad are deploying robots across hospitality, property management, and urban operations at i-City. The partnership plans to develop the world’s first AI and Robotics Residential Tower, positioning residential environments as a testbed for humanoid robot integration.

Also Read: Serving up the future: How robots are revolutionising the F&B industry

“This launch marks the first of several strategic initiatives we will roll out in the Asia-Pacific region throughout 2026 and beyond,” said Abel Deng, President, Asia-Pacific & Middle East Region, AGIBOT.

Tan Sri Lim Kim Hong, Chairman of I-Berhad, stated that the initiative aims to advance Malaysia’s positioning as a regional AI innovation hub and establish “a new benchmark for intelligent, human-centric living in the region.”

Product portfolio

AGIBOT’s commercial offerings span multiple use cases: the A2 series for reception and hospitality; the X2 series for entertainment and education; the G2 series for industrial manufacturing; the D1 series for inspection operations; and the C5 autonomous floor-care robot.

The company targets eight application areas, including reception, entertainment, industrial manufacturing, logistics, security, data collection, and research.

Regional context

AGIBOT’s expansion arrives at a pivotal moment for robotics adoption across Southeast Asia. While the region’s robotics sector remains smaller than those in Singapore or South Korea, growth is accelerating. Rising labour costs across manufacturing hubs like Malaysia, Thailand, and Vietnam are driving automation investments. Supply chain reshoring initiatives, accelerated by pandemic disruptions, are creating new opportunities for local robotics deployment.

The sector, however, faces challenges. Compared to mature markets, Southeast Asia has a smaller installed base of industrial robots, lower overall adoption rates in some industries, and a developing ecosystem of local robotics companies. Yet these barriers also represent opportunities for players like AGIBOT to shape market standards and build early-mover advantages.

Also Read: Robotics, space, sustainability: The forces shaping Asia’s next tech chapter

Key trends include increased AI integration, expansion beyond manufacturing into services, and growing emphasis on localised solutions tailored to regional needs.

Strategic trajectory

AGIBOT frames the Malaysia launch as the first of multiple 2026 initiatives. The company indicates plans to expand partnerships across Asia-Pacific and deploy robots in “closed-loop commercial scenarios”, operational environments where continuous learning and improvement can occur. The emphasis on robots-as-a-service models suggests a subscription-based approach rather than outright sales.

The post China’s humanoid robot leader AGIBOT sets sights on Southeast Asia appeared first on e27.

When Toku registered its offer document for a proposed listing on the Catalist Board of the Singapore Exchange, the announcement landed as more than just another IPO filing. For Singapore’s tech startup ecosystem, it signalled a potential inflexion point that touches public market confidence, regional-first tech building, the evolution of AI infrastructure and how startups scale across Asia.

At a time when global tech IPOs remain selective and private capital is increasingly disciplined, Toku’s move carries broader implications for founders and investors navigating 2026.

Perhaps the most immediate impact of Toku’s listing is narrative-driven rather than numerical. Singapore has only just begun to see the return of IPO activity among venture-backed tech companies in 2025. Against that backdrop, Toku’s decision to pursue a local listing reinforces the relevance of SGX Catalist as a viable exit route for growth-stage tech firms.

For founders, this matters. IPOs shape long-term decision-making well before a company reaches the listing stage. A credible example like Toku helps re-anchor expectations around governance, capital efficiency, and operational maturity — qualities that public markets now prioritise more than rapid, loss-driven expansion.

It also sends a message that startups do not necessarily need to look overseas for liquidity events. In an era of heightened geopolitical and regulatory complexity, a domestic listing pathway reduces execution risk while keeping companies anchored in Singapore’s ecosystem.

Also Read: Nasdaq tumbles, but Bitcoin soars past US$97K on massive short squeeze

Validating “Asia-built-for-Asia” tech

Toku’s positioning as an AI-powered customer experience platform, built specifically for complex, multilingual, and regulated markets, aligns closely with Singapore’s evolving tech identity. Rather than exporting standardised global products, the company has focused on solving problems that are particularly acute across the Asia Pacific, the Middle East, and emerging markets.

This reinforces a broader shift underway in the ecosystem: a growing emphasis on regionally grounded enterprise technology rather than consumer platforms or copycat SaaS models.

For investors, Toku’s IPO helps validate the commercial potential of startups that prioritise:

– Local compliance and data sovereignty
– Multilingual AI performance
– Deep integration with telecom and enterprise infrastructure

These are not easily replicable advantages, and they play to Singapore’s strengths as a regulatory-savvy, enterprise-facing tech hub. As a result, Toku’s listing could encourage more founders to pursue defensible, infrastructure-adjacent business models rather than chasing scale through uniform global deployment.

While AI remains a dominant theme across the startup landscape, much of the recent attention has focused on applications and user-facing tools. Toku’s story highlights a different layer of the AI stack: infrastructure that enables enterprises to deploy AI securely, compliantly and at scale.

Its end-to-end control over connectivity, orchestration and AI applications underscores the growing importance of systems-level innovation, particularly for regulated industries such as financial services, healthcare and government.

Also Read: Toku files for SGX Catalist IPO, doubles down on partner-led go-to-market strategy

This has implications for how investors assess AI startups in 2026. Rather than asking only what models or features a company offers, there is increasing scrutiny around how AI is deployed, governed, and integrated into existing workflows. Startups that can demonstrate robustness in these areas may find stronger traction with both enterprise buyers and capital providers.

Normalising partner-led go-to-market strategies

Equally significant is Toku’s emphasis on a partner-led go-to-market strategy as it scales beyond its core markets. Instead of relying solely on direct sales expansion, the company is doubling down on channel partners, systems integrators, and ecosystem alliances to efficiently enter new regions.

This approach reflects the realities of scaling in fragmented markets, where local knowledge, regulatory familiarity and on-the-ground execution often matter more than centralised sales teams. For other startups, Toku’s playbook offers a practical alternative to capital-intensive expansion models.

In 2026, this could influence how early- and growth-stage companies design their products and pricing with partners in mind from the outset. It also strengthens the role of Singapore-based enterprises, telcos, and consultancies as distribution enablers for regional startups.

More broadly, it suggests a maturing ecosystem where success is measured not only by the speed of expansion, but also by sustainable market entry and long-term customer value.

Also Read: The quiet currency shift: Southeast Asia’s strategic pivot to a multipolar monetary era

A signal beyond the listing

Taken together, Toku’s IPO represents more than a single company reaching the public markets. It reflects a broader recalibration underway across Singapore’s tech ecosystem, where investors and founders alike are placing renewed emphasis on fundamentals. In place of growth-at-all-costs narratives, there is a clearer preference for disciplined execution, clear paths to sustainability and tech that solves real, structural problems.

The listing also highlights a shift towards regional relevance over generic scale. Toku’s success has been built on addressing the complexities of multilingual, regulated and infrastructure-constrained markets, rather than forcing uniform solutions across geographies. This approach reflects a growing conviction that Singapore-based startups can succeed by embracing regional specificity, rather than competing directly with global incumbents on their own terms.

Equally important is what Toku’s story signals about the evolution of AI in enterprise settings. As excitement around AI matures, attention is shifting beyond surface-level applications to infrastructure, governance, and deployment at scale. Platforms that prioritise compliance, reliability and integration are increasingly viewed as more defensible and more valuable over the long term.

For founders, Toku’s IPO helps reset expectations around what success looks like in 2026. The pathway to public markets is no longer defined solely by speed, but by operational maturity, partner ecosystems, and credible regional strategies. These are traits that require time and patience to build, but which are increasingly rewarded.

While Toku’s long-term performance as a listed company remains to be seen, its decision to list locally already carries weight. It broadens the ambition set for Singapore’s tech sector, offering a concrete example of how companies can scale responsibly, remain regionally grounded and still access public market capital on home soil.

—

The lead image is AI-generated.

The post Why Toku’s public listing could reset expectations for Singapore startups appeared first on e27.

Asia-Pacific agrifoodtech startups raised US$4.2 billion in 2024, capturing 31 per cent of global sector funding. Yet across Southeast Asia, where agriculture remains a primary livelihood for hundreds of millions of people, this investment systematically misses the majority of farming communities. This isn’t just a market inefficiency—it’s a fundamental failure to understand how rural communities actually function.

Here’s the uncomfortable truth: every single agricultural app is designed for an imaginary user.

Picture Sari, a rice farmer in Central Java. Her smartphone buzzes with a weather alert suggesting she plant early because conditions are perfect. Sounds helpful, right? Except she can’t plant anything. Her village irrigation group hasn’t released water to her section yet. The planting schedule was decided three weeks ago in a community meeting she attended, but didn’t lead. Her father-in-law, the village elder, and the irrigation chief made that call together.

That weather app just became expensive digital noise.

This is the story of how Southeast Asian agritech is optimising for individual users in a world that makes collective decisions—and why the communities that need innovation most are getting left behind.

The fundamental misalignment

The person using your app isn’t the decision maker

The core problem isn’t technical sophistication or user experience design. It’s that the person using your app is rarely the person with power to change anything.

Consider the typical agritech user journey:

• Information flows to individuals (weather alerts, market prices, agronomic advice)
• Decisions are made collectively (planting schedules, input purchases, marketing choices)
• Impact requires community coordination (water management, pest control, harvest timing)

We’re building perfect steering wheels for passengers.

The three-layer reality of rural decision making

• Individual layer: The farmer who downloads your app, creates an account, receives notifications
• Community layer: Village councils, irrigation groups, farming cooperatives that make actual resource allocation decisions
• Ecosystem layer: Traders, input suppliers, financial institutions that control access to markets and credit

Most agritech stops at individual layer. Real agricultural change happens at community and ecosystem layer.

Also Read: Digital farming’s false promise: Why Asia’s US$180B bet on agritech-driven farming is failing smallholders

Case study: The US$15M weather app that changed nothing

AgriWeather (anonymised) raised US$15M Series A to provide hyperlocal weather forecasting to Indonesian farmers.

The pitch deck reality

• 250,000+ downloads in six months
• 85 per cent user retention after first month
• Sophisticated machine learning models
• Beautiful, intuitive interface in local languages

The field reality

In Bali’s rice terraces, they’d never heard of subak—thousand-year-old irrigation cooperatives that manage water allocation across entire watersheds. Planting schedules aren’t individual decisions; they’re collective negotiations about water timing, pest management, and harvest coordination.

When farmers showed the app’s recommendations to their subak leaders, the response was predictable: “We’ll plant when water reaches our section, like we discussed last month.”

The outcome

Great user metrics, zero agricultural impact. The startup eventually pivoted to B2B after burning through their Series A trying to acquire individual users who couldn’t act on their product.

The lesson

Technical sophistication is irrelevant if you’re solving the wrong problem for the wrong decision maker.

The collective decision-making framework

Understanding power structures across Southeast Asia

• Indonesia: Gotong royong (mutual cooperation) principles mean agricultural decisions flow through village councils (RT/RW) and traditional cooperatives (subak, gapoktan)
• Thailand: Farmer cooperatives and Royal-initiated agricultural groups maintain significant influence over individual farming practices
• Vietnam: Commune-level agricultural cooperatives (HTX) coordinate resource allocation and market access for member farmers
• Philippines: Barangay councils and irrigation associations (IA) make collective decisions about water management and crop timing

The four decision archetypes

• Water and timing decisions: Controlled by irrigation groups and village councils
• Input and credit decisions: Managed through cooperatives and traditional lending relationships
• Knowledge and technique decisions: Shared through farmer groups and extension networks
• Market and sales decisions: Constrained by existing trader relationships and cooperative marketing

Implication: Successful agritech must integrate with these existing structures, not bypass them.

Also Read: Indonesia’s agritech landscape: Keys to building a scalable agriculture startup

What actually works: The community-centred model

Case study: AgriCooperative’s US$50M success

Instead of targeting individual farmers, AgriCooperative worked through existing farmer cooperatives:

The approach

• Village leaders became part of the credit assessment process
• Community members provided social collateral
• The platform recognised that rural financial decisions involve community reputation

The results

• 1,847 farming families accessed US$12M in credit
• 94 per cent repayment rate (vs 67 per cent industry average for individual lending)
• Average household income increased 23 per cent over 18 months
• Platform became integral to existing community decision-making processes

Why it worked

They built for the reality of collective decision-making instead of fighting it.

The new framework: Community-centred agritech

Design principles for rural reality

• Map Power Before Building Product

• Who actually makes agricultural decisions in target communities?
• How do information and resources flow through village structures?
• What existing institutions could be enhanced rather than replaced?

• Design for Collective Intelligence

• Individual apps that feed into group decision-making processes
• Platforms that strengthen existing community coordination mechanisms
• Tools that make collective decision-making more efficient, not obsolete

• Measure Community-Level Impact

• Agricultural outcomes across entire villages or cooperatives
• Strengthening of traditional knowledge and decision-making systems
• Economic improvements at the household and community level

• Build Sustainable Revenue Models

• Community-centred solutions often require different monetisation approaches
• Success may come through institutional partnerships rather than individual payments
• Value creation happens at the ecosystem level, not just user level

Also Read: How Southeast Asia’s agritech startups are turning smallholder farms into high-tech powerhouses

Implementation roadmap

Phase one: Deep community research (three-six months)

• Ethnographic study of agricultural decision-making in target regions
• Mapping of existing community institutions and power structures
• Identification of technology integration points within collective systems

Phase two: Community partnership development (6-12 months)

• Build relationships with village leaders and cooperative managers
• Co-design solutions with community decision-makers
• Pilot programs that enhance existing coordination mechanisms

Phase three: Scaled community integration (12+ months)

• Deploy solutions through established community channels
• Measure impact at village and cooperative level
• Refine based on community-level feedback and outcomes

The investment challenge and opportunity

Why VCs struggle with community-centred models

• Traditional VC Metrics:

• Individual user acquisition costs
• Monthly active users
• Revenue per user
• Individual customer lifetime value

• Community-centred Metrics:

• Community adoption rates
• Collective behaviour change
• Village-level agricultural outcomes
• Ecosystem-wide economic impact

The multi-billion dollar opportunity

• Current market focus: Southeast Asian agritech primarily serves educated, connected, individually empowered farmers—a small fraction of the total agricultural population
• Underserved market: The vast majority of rural populations making collective decisions through traditional institutions like cooperatives, irrigation groups, and village councils
• Market potential: ASEAN agricultural production alone totaled 195.5 million tons of rice, 44.7 million tons of corn, and 82.1 million tons of cassava in 2022, representing massive economic activity largely untouched by current agritech solutions

New investment framework

• Community traction metrics:

• Number of village councils or cooperatives actively using the platform
• Collective decisions influenced by platform insights
• Community-level agricultural outcome improvements
• Integration depth with existing institutional structures

• Revenue model innovation:

• Institutional partnerships with cooperatives and government extension services
• Value-based pricing tied to community-level outcomes
• Revenue sharing with traditional institutions rather than competing with them

The competitive advantage of getting this right

Companies that successfully engage rural communities through collective decision-making systems will access currently underserved agricultural markets while building sustainable competitive advantages through deep community integration.

• Policy alignment opportunity

Asian Development Bank research highlights that Southeast Asian governments prioritize strengthening rural statistical systems and agricultural coordination—community-centered solutions align with these policy goals while improving agricultural outcomes.

• Investment trends supporting community-centred models

The 38 per cent year-over-year increase in Asia-Pacific agrifoodtech funding demonstrates growing investor confidence in agricultural innovation, creating opportunity for solutions that demonstrate genuine community-level impact rather than just individual user metrics.

Also Read: Need of the hour: How agritech platforms can protect farmers from climate change

The choice point for Southeast Asian agritech

• The individual path (Current trajectory)

• Target educated, connected farmers representing a small fraction of the agricultural population
• Compete on user experience and technical features
• Measure success through app-store metrics and individual user engagement
• Generate venture returns through individual customer acquisition in urban-adjacent markets

Outcome: Sophisticated solutions serving a narrow market segment while the majority of agricultural communities remain underserved

• The community path (Untapped opportunity)

• Partner with existing rural institutions like cooperatives and village councils
• Enhance collective decision-making processes that govern resource allocation
• Measure success through community-level agricultural outcomes and livelihood improvements
• Generate returns through ecosystem-wide value creation and institutional partnerships

Outcome: Access to vast underserved agricultural markets while creating genuine rural development impact

Call to action: Building for rural reality

• For entrepreneurs: Stop building for imaginary individual farmers. Start with deep community research. Map power structures before writing code. Design for collective intelligence rather than individual optimization.
• For investors: Fund founders who understand rural power dynamics. Demand community-level impact metrics before major funding rounds. Recognise that community-centred solutions may have different growth curves but potentially larger ultimate markets.
• For the ecosystem: Question who actually benefits from “agricultural innovation.” If the answer is primarily urban, tech-savvy users, we’re solving the wrong problems with the wrong metrics.

The agricultural community question

Southeast Asian agritech has attracted billions in investment to serve the region’s agricultural communities. Yet most of those communities remain unreached by the innovation they’re funding.

The technology exists. The market need is massive—ASEAN’s agricultural output continues growing, with rice production forecast to increase to 202.34 million tons in 2024. The missing piece is understanding that rural communities don’t function like individual app users—and that’s not a problem to solve but a reality to build for.

Rural Southeast Asia is ready for agricultural innovation. The question is whether the 31 per cent of global agrifoodtech funding flowing to Asia-Pacific will finally reach the communities that actually exist, not the individuals we imagine.

The communities are waiting. What are we going to build for them?

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. Share your opinion by submitting an article, video, podcast, or infographic.

Enjoyed this read? Don’t miss out on the next insight. Join our WhatsApp channel for real-time drops.

Image courtesy: Canva Pro

The post The rural-urban innovation divide: Why billions in agritech investment is missing most of Southeast Asia appeared first on e27.

Over the past few years, AI in the enterprise has been about trials and piloting of new solutions. While over 70 per cent of banking institutions globally use agentic AI today, only 16 per cent have moved to actual operational deployment. This gap reveals a critical truth: the barrier to production scale is not technology. It is organisational clarity and strategic intent.

In 2026, as we move beyond AI hype to reality, executives across financial services, insurance, healthcare, and technology will face a decisive question: are we committing to production-scale deployment, or continuing to cycle through experiments?

The upside for organisations is substantial. McKinsey estimates AI could unlock US$340 billion in annual value for banking alone. Yet less than one per cent of global executives report significant ROI from AI investments, defined as 20 per cent or greater improvement in profitability or cost savings. Only three per cent report a substantial ROI of 10 to 20 per cent. This disparity reflects a systematic failure of execution.

Across industries, we see that when institutions successfully transition from pilots to production scale, they achieve improvements in processing speed, automation rates, and regulatory compliance readiness.

The strategic shift: From cost reduction to autonomous efficiency

Organisations that are succeeding at production scale are not optimising for headcount reduction, but instead, they are optimising for autonomous efficiency. This means they are using AI to eliminate routine work completely so humans can focus on revenue-generating activities.

For example, this could include a lending algorithm approving customer applications instantly without human review or a fraud detection system blocking suspicious transactions automatically. Proven revenue levers include hyper-personalised cross-sell and upsell, financial inclusion lending powered by alternative data, premium digital wealth management, and AI-augmented compliance and fraud prevention.

Also Read: Looking beyond the bots: The unsexy digital skills that actually matter in 2026

This reframing of autonomous efficiency changes ROI calculations fundamentally. According to Larridin’s 2025 Enterprise AI Report, enterprises measuring AI properly report average productivity improvements of 27 per cent, time savings of 11.4 hours per knowledge worker weekly, and cost reductions of US$8,700 per employee annually. But these improvements accrue only to organisations that commit to autonomous workflows rather than basic human augmentation layers.

For example, the production-ready approach Dyna.Ai has pioneered demonstrates this principle. By operating at sub-200 millisecond response times with over 95 per cent accuracy, the platform enables businesses to deploy autonomous agents across lending decisioning, fraud detection, and customer engagement workflows. These are not experimental applications. They are production systems handling millions of transactions with consistent, measurable performance.

Southeast Asia is laying the foundations for enterprise AI success

Singapore has long been an innovation testing ground, and now it is at a unique inflection point. With the National AI Strategy 2.0, fostering nearly 900 AI startups and attracting US$1.04 billion in fintech investment in H1 2025 alone, the country and the region are establishing themselves as a production hub for enterprise AI. Unlike markets where AI remains primarily experimental, Southeast Asia is seeing financial institutions move directly from exploration to execution. 

Southeast Asian banks are doubling AI-driven value. For instance, one bank grew from US$273 million to US$555 million year-over-year, while DBS generated US$565 million from 350 use cases in 2024 through RM co-pilots and inclusion lending. 

The acceleration of AI adoption reflects both market dynamics and regulatory clarity. The Monetary Authority of Singapore’s sandbox approach and ASEAN regulatory frameworks are creating conditions where institutions can deploy AI at scale with defined governance structures.  

Also Read: 4 marketing trends that will dominate budgets and strategies in 2026

The C-Suite must lead AI governance, infrastructure, and delivery

Moving from pilot to production requires establishing robust governance frameworks before scaling. This means creating detailed AI system inventories, prioritising use cases with clear business value, ensuring data quality, and designing workflows to enable AI autonomy with human oversight. Production success hinges on delivering AI directly into workflows like RM consoles and mobile apps, implementing policy-as-code governance, and pursuing smart partnerships: buy to explore, partner to scale using embedded squads, API-first integration, and revenue-linked contracts.

According to the Bank of England’s Artificial Intelligence in UK Financial Services survey, 84 per cent of firms have established accountable persons for their AI frameworks, and 72 per cent allocate accountability for AI use cases to executive leadership. Yet the same survey reveals that 46 per cent of firms report only partial understanding of the AI technologies they deploy, particularly those sourced from third parties. This gap between governance structures and technical understanding underscores why production-scale deployment requires simultaneous investment in governance clarity, data modernisation, and organisational capability-building.​

Data infrastructure matters equally. Financial institutions increasingly recognise that real-time data capabilities and robust data governance are foundational to production-scale AI deployment. Organisations that establish this infrastructure early, alongside accountability structures and technical understanding of their AI systems, will execute production deployment faster than those attempting simultaneous infrastructure modernisation and AI scaling.

The production transition won’t happen immediately, and organisations must establish foundational infrastructure first, operationalise early wins with measurable business metrics, then scale from demonstrated success. Those executing this roadmap rigorously will move from being stuck in AI pilots to achieving actual production scale. 

With AI adoption growing across sectors, the demand for solutions is evident. What remains is organisational will. In 2026 and beyond, that will be the differentiator between leaders and laggards across financial services, insurance, technology, and beyond.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. Share your opinion by submitting an article, video, podcast, or infographic.

Enjoyed this read? Don’t miss out on the next insight. Join our WhatsApp channel for real-time drops.

Image courtesy: Canva

The post AI’s tipping point: Why 2026 will separate the leaders from the laggards in financial services appeared first on e27.

There’s a paradox at the heart of modern international business.

We’ve digitised nearly every aspect of operations. Marketing runs on sophisticated attribution models. Sales teams use AI-powered CRMs. Supply chains are optimised down to the last mile.

Yet when it comes to moving money across borders—one of the most fundamental business functions—most companies are still using infrastructure designed for a pre-digital era. And paying dearly for it.

The opacity problem in cross-border payments

Traditional banking thrives on information asymmetry.

When you execute an international wire transfer through a legacy bank, you’re operating in a black box. The exchange rate you receive isn’t the interbank rate—the actual rate at which currencies trade. It’s a marked-up version, typically 1.5-2% above market.

This markup doesn’t appear as a line item. It’s embedded in the rate itself. Your finance team sees money leave in one currency and arrive in another, with what appears to be normal conversion loss.

The real question is: how much of that “conversion loss” is actual market movement, and how much is bank margin?

Most companies can’t answer that question. And that’s precisely the point.

The economics of scale—working against you

Here’s what makes this particularly painful for growing businesses.

As you scale internationally, this problem compounds. A startup processing $100K in cross-border payments loses $1,500-$2,000 annually to FX markups—annoying, but not existential.

A mid-market company processing $10M? That’s $150,000-$200,000. Suddenly, hidden FX fees rival your entire marketing budget.

Yet finance teams often lack visibility into this cost center because it’s not categorized as “fees”—it’s just absorbed into COGS or operational expenses.

This is margin erosion in its purest form: value quietly transferred from your business to financial intermediaries, with no corresponding increase in service quality.

Also read: Why WorldFirst’s latest move could change how digital platforms scale worldwide

The infrastructure shift that’s already happening

Forward-thinking CFOs have started asking a different question.

Not “What does our bank charge?” but rather “What does international payment infrastructure actually cost in 2025?”

The answer increasingly involves purpose-built fintech platforms designed for cross-border commerce. Companies like WorldFirst have built infrastructure that:

1. Provides transparent, near-market FX rates—typically 0.3-0.5% above interbank, versus 1.5-2% for traditional banks
2. Eliminates hidden intermediary fees—no correspondent banking markups
3. Offers real-time visibility—you know exactly what you’re paying before you commit

This isn’t about chasing marginal savings. It’s about fundamentally rethinking payment infrastructure as a strategic function rather than a commodity service.

The new customer economics

WorldFirst’s current offer for new customers illustrates where the market is heading:

• 5 free international transfers (complete fee waiver)
• 50% reduction in FX costs on ongoing transactions
• Auto-applied incentives (no manual intervention required)

For a business processing $1M annually in cross-border payments, this translates to $10,000-$15,000 in recovered margin, every year, indefinitely.

But the more interesting insight isn’t the promotional pricing. It’s that platforms can afford to offer these economics because their infrastructure costs are fundamentally lower than legacy banking systems.

This is what disruption actually looks like: not flashy innovation, but structural cost advantages that incumbents can’t match without cannibalizing their existing business model.

What this means for strategic planning

If you’re a CFO or finance leader planning for 2025, cross-border payment optimisation should be on your roadmap—not as a “nice to have,” but as a margin protection initiative.

Ask yourself:

• Do we have visibility into our actual FX costs? Not what we’re told they are, but what the market spread actually is?
• Are we treating payment infrastructure as strategic or commodity? If commodity, you’re likely overpaying.
• What would we do with an extra $15K, $50K, or $150K in annual margin? Because that’s what’s at stake.

The pattern we’re seeing

We work with hundreds of scaling businesses across Southeast Asia and beyond. The pattern is unmistakable:

Companies that successfully scale internationally have optimized their payment infrastructure early.

Not because they’re obsessed with saving pennies. But because when you’re operating on 10-20% margins (typical for e-commerce, manufacturing, or marketplace businesses), a 1.5% hidden cost is the difference between sustainable growth and a slow bleed.

The businesses still using traditional banks for cross-border payments fall into two categories:

1. Early-stage companies that haven’t reached scale where it matters yet
2. Mid-market companies with institutional inertia—”we’ve always done it this way”

The latter group is leaving significant money on the table. And in an increasingly competitive global market, that’s a luxury fewer businesses can afford

Also read: What facilitates the adoption of digital currencies in Southeast Asia?.

The action item

This isn’t a dramatic transformation. You don’t need to rip out existing systems or retrain your entire finance team.

It’s a simple evaluation:

1. Calculate your actual annual cross-border payment volume
2. Estimate your current all-in FX costs (including hidden markups)
3. Compare to alternative infrastructure pricing
4. Run a pilot with a portion of your volume

For most businesses with meaningful international exposure, the ROI is immediate and substantial.

The question isn’t whether to optimize payment infrastructure. It’s whether you can afford not to.

Ready to benchmark your current FX costs? Explore WorldFirst’s new customer offer and see where you stand.

—

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

The e27 team produced this article

We can share your story at e27 too! Engage the Southeast Asian tech ecosystem by bringing your story to the world. You can reach out to us here to get started.

Featured Image Credit: Canva Images

The post The hidden margin killer: Why CFOs are rethinking cross-border payment infrastructure appeared first on e27.

For much of the internet’s history, visibility operated like a formula. Feed any search engine the right signals — keywords, backlinks, structured data — and your content could rise through the search rankings to be seen by potential customers. Digital influence was largely a technical exercise, but the way people seek information has fundamentally changed.  Generative AI systems are now used to remove the labour of decision-making and are expected to produce answers and recommendations, not search results.

This shift is reshaping the entire logic of discoverability. When answers replace links, the criteria for visibility changes too. That is where Generative Engine Optimisation (GEO) comes in: the emerging discipline that determines how brands appear in AI-generated responses or whether they appear at all.

From signals to semantics: A new foundation for visibility

Traditional SEO rewards pages that satisfy ranking algorithms. GEO, however, is rooted in meaning. Large language models (LLMs) also index content, but that’s not all they do.  It interprets data for users, it sets criteria, it makes recommendations, and when challenged, it doubles down on those recommendations.

In this world where LLMs mediate decisions, visibility is earned through consistency. LLMs look for and evaluate coherence, narrative alignment, and reliability across the broader information ecosystem when generating responses or recommendations.  To appear, a brand must sound consistent no matter where it appears: an owned website, a media interview, an annual report or an analyst brief. When generative models detect stable patterns, they treat them as trustworthy reference points. Conversely, when they encounter contradictions, they simply omit the brand from the answer.

The question for leaders shifts from “How do we rank?” to “How are we described when we are not in the room?”

The new hierarchy of credibility

GEO reshapes traditional communication principles by redefining what credibility looks like in an AI-mediated landscape. Experience is no longer about broad claims but about demonstrable outcomes and evidence of impact. Expertise is conveyed through spokespersons whose perspectives are clear, quotable, and consistent.

Also Read: AI in banking: Unlocking success with ChatGPT and embracing the future

Authority stems from being featured in reputable, high-quality platforms and contexts that AI systems recognise as reliable, like events, in traditional media. Lastly, trust emerges when a brand’s internal messaging aligns with how external sources describe it. Together, these elements create a semantic identity: a coherent, machine-readable portrait of the brand.

AI as the new gateway to information

As generative engines take over more search behaviour, the cost of inconsistency grows. A brand that doesn’t appear in AI-generated answers becomes digitally invisible, even if its SEO footprint remains strong. Meanwhile, companies with aligned narratives gain semantic weight and become the default examples referenced by LLMs.

We are already seeing early adopters shape how entire industries are defined. Their language becomes the vocabulary that AI uses to describe the market.

Transforming communications strategies

GEO also changes how communications leaders create and monitor content. Content should not be viewed as a standalone asset; it is crucial data input that AI systems analyse and learn from. This makes structure as important as storytelling, demanding content that is precise, contextual, and easy for machines to interpret. Credible media placement gains new weight as LLMs increasingly prioritise trusted sources.

At the same time, monitoring now extends beyond sentiment or volume to assessing how AI systems describe the brand, what they overlook, and where misunderstandings occur. Ultimately, influence is shifting from optimising for algorithms to optimising for the quality and accuracy of the answers machines produce.

GEO is not a replacement for SEO — it is the next layer

While SEO ensures content is accessible, GEO ensures it is understood, making both essential for modern visibility. To succeed in this new environment, brands must regularly audit how they appear across generative systems, address narrative inconsistencies across channels, and create content that is reusable, structured, and easily quotable. It also requires treating communication as a unified ecosystem rather than a collection of isolated outputs. In this context, meaning and distribution — not volume — becomes the decisive asset.

Also Read: Are large Vietnamese tech enterprises ‘indifferent’ when competing with ChatGPT?

What this means in the new year

Generative AI will no longer be a novelty but the main interface through which information is accessed. Search will feel less like “searching” and more like conversing. Consumers will expect direct, personalised answers, and brands will compete for inclusion within those answers, not for page-one rankings.

GEO will become a baseline requirement for digital existence. Brands that invest early in semantic clarity and consistency will shape category narratives. Those that lag may find themselves gradually omitted from the AI-generated knowledge graph — a form of invisibility that is difficult to reverse once established.

The organisations best prepared for this future understand one thing clearly: in the age of AI, visibility is not determined by how loudly you speak, but by how clearly you are understood.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. Share your opinion by submitting an article, video, podcast, or infographic.

Enjoyed this read? Don’t miss out on the next insight. Join our WhatsApp channel for real-time drops.

Image credit: Canva

The post Do you know what ChatGPT is saying behind your back? appeared first on e27.

For a long time, venture studios were defined by their ability to build companies end-to-end. A single team would generate ideas, form founding teams, develop products, and support early fundraising. For several years, this integrated model worked well.

As we move through 2025 and prepare for 2026, however, that definition is starting to fall short. Across Asia and other startup ecosystems, a clear pattern is emerging: venture studios are increasingly partnering with one another, and this is less a trend than a structural adjustment to how startups are now built and scaled.

Why the single-studio model is no longer sufficient

The operating environment for early-stage companies has changed in meaningful ways.

• First, global readiness is expected much earlier. Local validation alone is rarely enough to support long-term growth or follow-on investment. Market entry strategy, early partnerships, and initial sales conversations now need to be considered from the start.
• Second, execution capabilities have become more specialised. Product development, go-to-market execution, partnership building, and fundraising each require distinct skill sets. Maintaining excellence across all of these functions within a single studio has become increasingly inefficient.
• Third, precision matters more than speed. Starting quickly is no longer the main advantage. What matters more is being connected to the right customers, partners, and investors at the right moment.

In response, venture studios are rethinking where they create the most value—and where collaboration makes more sense than internal ownership.

Also Read: Real estate sales development: Unlock the power of partnership and collaboration

Collaboration as role clarity, not expansion

The recent rise in venture studio partnerships is often misunderstood as an effort to scale faster or increase visibility. In practice, most collaborations are far more deliberate. They are based on clear functional role-sharing rather than broad cooperation.

Some studios focus on early company formation and business design. Others specialise in market entry, sales execution, or cross-border partnerships. Still others are strongest in capital formation and investor networks. Increasingly, studios are choosing not to duplicate these capabilities internally.

Recent partnerships, including collaborations with firms such as One Tree Hill Ventures, reflect this approach. Rather than attempting to control the entire startup lifecycle, each organisation focuses on the stage where it can operate most effectively. Outcomes are then passed to the next execution partner in a structured way. The objective is not speed for its own sake, but reducing execution risk and building repeatable paths to growth.

Addressing the execution gap after introductions

Another factor driving collaboration is a persistent gap in the startup ecosystem: strong initial engagement, weak follow-through.

Demo days, conferences, and curated meetings have multiplied, yet many promising conversations fail to translate into concrete outcomes. This challenge is not limited to founders. Venture studios and accelerators face the same issue internally, where introductions are made but ownership of next steps remains unclear.

Partnership-driven models help address this problem by clarifying responsibility. When execution roles are explicitly defined across organisations, connections are more likely to move beyond discussion and toward action. In this sense, collaboration becomes less about expanding networks and more about increasing execution density.

Redefining venture studio competitiveness for 2026

As we approach 2026, venture studio performance is no longer judged primarily by how many startups are launched or how quickly ideas are turned into products.

Instead, more relevant questions are emerging:

• Where does this organisation create the highest execution leverage?
• Which parts of the startup journey are better handled by partners?
• Can this structure be repeated and scaled across multiple companies?

Also Read: Weathering the tariff turbulence: How AI and collaboration can lift SEA SMEs

Partnerships offer practical answers to these questions. They are not a signal of weakness, but a recognition that specialised strengths, when properly connected, outperform fully integrated but diluted models.

Collaboration as a sign of maturity

The growing number of venture studio partnerships suggests that the sector itself is maturing. Organisations are becoming more explicit about what they do well—and equally clear about what they choose not to do.

Looking ahead to 2026, the differentiator for venture studios will be less about how much they can build alone and more about how clearly they define roles, connect execution, and sustain those structures over time.

Collaboration, in this context, is not a compromise. It is a strategic response to a more complex and interconnected startup environment.

—

The post Why venture studios are choosing collaboration over competition appeared first on e27.

If you walk around the Gelora Bung Karno (GBK) stadium complex in Jakarta on a Sunday morning, you will witness a fascinating spectacle. It is a runway of neon-colored carbon-plated shoes, smartwatches that cost more than a motorcycle, and activewear that screams luxury. Running in Indonesia’s capital—and in other major hubs like Surabaya or Malang—has transcended mere cardio. It has become the supreme social currency of the urban middle class.

But recently, a glitch has appeared in this well-curated matrix.

A new, bizarre service has surfaced in the underbelly of X (formerly Twitter) and community Telegram groups, creating ripples of confusion and amusement across the local tech ecosystem. They call themselves “Joki Strava” (Strava Jockeys).

For a fee ranging from IDR 50,000 to IDR 100,000 (roughly US$3 to US$6), these individuals offer a service that sounds like a plotline from a dystopian satire: they will log into your fitness account, strap your phone (or theirs) to their arm, and run a 10K at a blistering pace on your behalf.

You get the stats. You get the glamorous map route. You get the kudos. They get the sweat.

As a tech observer based in Indonesia, I find this phenomenon to be more than just a quirky viral trend. It is a profound case study on the fluidity of Southeast Asia’s gig economy, the commodification of data, and the extreme lengths users will go to purely for digital validation.

The mechanics of ‘outsourced’ vitality

To understand the Strava Jockey, one must first understand the unique digital landscape of Indonesia. This is a country where the informal economy has always been incredibly agile in adapting to digital platforms. We have seen “click farms” selling likes, “game jockeys” ranking up Mobile Legends accounts, and now, we have fitness proxies.

The transaction is shockingly simple, bypassing the need for complex APIs or platform loopholes. It relies entirely on crude account sharing—a cybersecurity nightmare, yet a risk users are willing to take. The client provides their login credentials. The jockey, often a genuine athlete or a student with high stamina and low cash flow, performs the activity.

Once the run is complete, the data syncs. The client then screenshots the “Morning Run” summary—complete with an impressive Pace four (four minutes per kilometre) and a high calorie burn—and posts it to Instagram Stories.

The caption usually involves faux-humility: “Felt heavy today, but glad I pushed through.” Meanwhile, the actual runner is likely catching their breath on a curb, waiting for the bank transfer to arrive.

Also Read: Indonesia’s antivirus reliance: A cybersecurity blindspot

Why buy sweat? The economy of vanity

From a Silicon Valley perspective, this makes zero sense. The value proposition of Strava is self-quantification; cheating defeats the entire purpose of the product.

However, from a Southeast Asian sociological perspective, it makes perfect sense. In Jakarta’s hyper-competitive social hierarchy, health is the new luxury. Being fit signals that you have the time and discipline to train—assets that are scarce in a city known for its punishing work hours and gridlock traffic.

A Strava screenshot is not just data; it is a “Proof of Life” for the elite. It signals: “I am part of the successful tribe.”

The demand for jockeys arises from a gap between aspiration and reality. The peer pressure to join running clubs (which are essentially networking hubs) is immense. But building the aerobic base to run a 10K takes months of painful effort. The vanity economy offers a shortcut: Buy the result, fake the process.

It is the digital equivalent of wearing a knock-off Rolex. The function is irrelevant; the signalling is everything.

The resilience of the micro-gig economy

The Strava Jockey phenomenon offers a crucial insight into the Indonesian market: If a platform has a social metric, locals will find a way to monetise it.

We often talk about the “Gig Economy” in the context of Gojek or Grab—formalised, app-based labour. But the Strava Jockey represents the “Shadow Gig Economy.” It is unregulated, decentralised, and incredibly efficient.

These jockeys are micro-entrepreneurs. They have identified a market inefficiency (rich people want stats but hate running) and provided a solution. They are monetising their own biological assets (lungs and legs) in a direct peer-to-peer transaction.

It also highlights a form of “Platform Leakage.” The transaction happens off-platform (negotiated on WhatsApp, paid via QRIS/e-wallet), meaning Strava captures none of the value, even though their app is the core product being sold.

Also Read: Malaysia, Indonesia escalate AI oversight with temporary Grok block

A challenge for health-tech trust

While amusing, this trend poses a serious question for the future of health-tech and insurance-tech (insurtech) in the region.

As insurance companies increasingly move towards “wellness-based pricing”—offering lower premiums to users who share their fitness data—the existence of Strava Jockeys breaks the trust model. If a user can outsource their cardio to a semi-pro runner, the data becomes corrupted.

How can an algorithm differentiate between a 40-year-old corporate executive suddenly running a sub-40-minute 10K, and a 20-year-old jockey carrying his phone?

Conclusion: The black mirror of the tropics

The rise of the Strava Jockey is a quintessentially Indonesian tech story. It blends high-tech adoption with deep-seated cultural behaviours—specifically panjat sosial (social climbing) and gotong royong (mutual assistance, even in cheating).

It serves as a reminder to founders and investors targeting this region: You can build the most sophisticated tracking algorithm in the world, but you cannot code against human nature.

In the vanity economy, reality is negotiable. We have entered an era where your Uber driver can bring you food, your Gojek driver can bring you packages, and now, your Strava Jockey can bring you health—or at least, the digital illusion of it.

The sweat is real. The stats are real. The only fake thing is the person claiming the glory. And in the economy of likes, perhaps that is the only metric that matters.

—

The post The rise of ‘Strava Jockeys’: How Indonesia’s vanity economy is hacking the fitness tech ecosystem appeared first on e27.

When my mother retired, she still loved cutting hair for seniors at the community centre. Her old clipper became her favourite companion, a symbol of care and pride. Years later, as her memory began to fade, she could no longer find that clipper. She searched everywhere, frustrated and sad, certain it was still around.

I realised then that she was not just looking for an object. She was searching for a piece of herself, the part that gave her purpose.

That moment became the start of my journey with AI storytelling. Because memory, once lost, is hard to retrieve. But the story, when shared, becomes timeless.

Preserving what fades

When a loved one begins to forget, their stories start to scatter like leaves in the wind. AI tools today can help us gather those fragments and hold them gently.

I used ChatGPT to help me write down her memories. I used Artflow to recreate images of her when she worked, laughing with her clients. I added her voice using simple audio tools. Suddenly, I had something precious, not just data, but emotion captured in motion.

It was not perfect, but it was deeply human. AI did not replace her story. It helped me remember it.

When technology becomes empathy

The truth is, AI cannot feel. But it can help us feel more. It can listen patiently, arrange words and images, and remind us of details we might overlook.

For families facing dementia, this becomes powerful. When you turn daily conversations into short stories, photos into memories, and voices into keepsakes, you are not using technology. You are using love in a new language.

Every time I create a short AI film about my mother, I feel as though I am giving her story back to her. It is a conversation between past and present.

Also Read: Preserving memories in the age of AI: How technology helps us remember who we are

Storytelling as connection

AI storytelling is not only for families. It can help communities preserve culture, educators record wisdom, and midlifers document their second acts.

We often underestimate the stories we carry. But every memory, even an ordinary one, can spark belonging.

When someone says, “No one wants to hear my story,” I remind them that memory is not about the audience. It is about continuity. It is how we remind ourselves that we mattered, that we made someone smile, that we once changed a small part of the world.

Healing through creation

The act of turning memories into art is healing in itself. You do not need technical skills. You only need intention.

When I guide participants in storytelling workshops, they often cry, laugh, or sit in quiet reflection. AI becomes a mirror for emotions they did not know how to express. Some use it to honour a lost parent. Others use it to capture childhood laughter or forgotten dreams.

The process heals because it allows us to hold both the pain and the beauty of remembering.

The gentle reminder

Stories are our real inheritance. They carry the colours of who we are. AI simply gives us a new brush to paint them with.

So if you have a memory worth keeping, do not wait. Speak it. Record it. Write it. Let AI help you shape it, not to make it perfect but to make it last.

Because one day, someone you love will look for a piece of you the same way my mother looked for her clipper. And when they do, your story will be there, waiting to be found.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. Share your opinion by submitting an article, video, podcast, or infographic.

Enjoyed this read? Don’t miss out on the next insight. Join our WhatsApp channel for real-time drops.

Image credit: Canva

The post AI storytelling for healing: Turning memories into digital legacies appeared first on e27.

In 2025, artificial intelligence (AI) transitioned from a boardroom buzzword to a primary driver of unemployment. The shift was surgical, targeting specific job functions that were once considered the bedrock of corporate tech.

According to data compiled by UK-based forex company RationalFX, nearly 245,000 jobs were lost in 2025 as companies swapped human salaries for software subscriptions.

Customer support: The first domino

The most visible victim of this transition was the customer support sector. Salesforce, a leader in CRM software, provided a stark example of this trend. CEO Marc Benioff announced that the company had slashed its customer support workforce nearly in half (from 9,000 positions to just 5,000) by deploying AI agents last year.

Also Read: Big Tech’s efficiency paradox: Record profits, record layoffs

Similarly, Amazon confirmed 14,000 job cuts specifically linked to AI adoption and the goal of making its corporate structure “leaner”. These 14,000 roles were part of a larger 20,000-person layoff wave at the company in 2025.

Amazon’s leadership described AI as “the most transformative technology we’ve seen since the Internet.” However, for thousands of employees in customer service and HR, that transformation meant the end of their livelihoods.

Replacing the “paper pushers”

The automation wave is also sweeping through HR, marketing, and financial operations. IBM, one of the industry’s oldest players, cut roughly 9,000 roles in 2025. While the company was tight-lipped about the specifics, reports indicated that the restructuring focused on non-tech jobs that could be partially replaced by AI. IBM successfully automated routine tasks such as drafting emails, managing internal queries, and analysing spreadsheets.

Professional services giant Accenture is executing a similar strategy. The firm announced a sweeping reduction of 11,000 employees over just three months as part of a US$865 million restructuring plan to pivot toward AI-driven operations. Even as it lays off thousands, it is doubling its AI and data specialist headcount, which now stands at 77,000.

The rise of “AI-first” hiring

The report by RationalFX highlights a disturbing trend: even when companies were not actively laying off, they were refusing to hire humans for roles that an AI can perform. Duolingo CEO Luis von Ahn clarified that while the company would not necessarily fire existing staff, they would only hire a human if “the AI cannot do the job it is tasked with properly”. This “AI-first” hiring policy has already led to the elimination of hundreds of contractor positions at the language-learning app.

Also Read: Why Asia’s tech giants are cutting from the middle

As enter 2026, the “middle class” of tech — support, administration, and middle management — finds itself in the crosshairs of a technology that doesn’t sleep, doesn’t require benefits, and, increasingly, doesn’t make mistakes.

The post AI’s first real casualties: The tech jobs that vanished in 2025 appeared first on e27.