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

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

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

The financial terms of the Trustana acquisition were not disclosed.

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

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

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

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

Why product data matters

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

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

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

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

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

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

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

A broader customer base

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

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

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

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

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

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

The competitive field

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

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

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

Singapore’s AI commerce play

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

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

Also Read: Revolutionising retail: A blueprint for future success

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

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

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

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

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

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

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

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

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

Education and government remain in the firing line

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

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

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

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

APAC remains one of the world’s most attacked regions

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

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

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

Also Read: What AI safety researchers actually worry about

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

GenAI turns into a daily data risk

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

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

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

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

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

Email remains the old reliable route for attackers

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

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

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

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

Ransomware breaks from its earlier pattern

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

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

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

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

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

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

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

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

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

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

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

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

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

The AI budget is still rising

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

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

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

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

That is where many finance teams are getting stuck.

Fragmented data is the real bottleneck

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

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

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

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

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

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

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

Singapore is ahead, but not fully autonomous

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

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

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

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

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

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

Talent becomes part of the infrastructure

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

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

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

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

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

What this means for Southeast Asian companies

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

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

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

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

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

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

Few words in business are abused as generously as ecosystem.

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

Most of the time, it is not.

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

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

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

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

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

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

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

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

Feature breadth is not the same as market orchestration

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

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

Perhaps. But perhaps not.

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

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

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

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

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

Real platforms create politics, fake ones create packaging

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

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

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

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

Many ecosystem claims are really attempts to borrow strategic prestige

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

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

So firms start narrating themselves upwards.

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

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

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

That tension usually sits at the heart of the illusion.

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

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

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

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

They are not.

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

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

—

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