Posted on — Leave a comment

Michael Padilla: The costliest failure is refusing to stop when strategy no longer holds

Michael J. Padilla

In a world where former military officers often trade uniforms for predictable government consultancies, Michael J. Padilla chose a different path. After more than 31 years in the US Army, rising to the rank of colonel, he founded Al Thuraya Holdings, a global group that today spans 18 companies across risk management, security, technology, consulting, energy, and business services.

What distinguishes Padilla isn’t merely the scale of what he built, but the hard-earned experience behind it: decades of leading through uncertainty, operating in high-risk environments, and helping governments and multinational corporations navigate a world that rarely behaves the way strategy documents predict.

The origins and logic of Al Thuraya

Padilla’s decision to leave the military wasn’t driven by a single dramatic moment but by a slow accumulation of observations. Over three decades, he watched standards shift, organisations grow more political, and coalition partners pursue their own competing interests.

Also Read: The 3Cs+1 framework: Navigating geopolitical fragmentation as a founder

He understood discipline, planning, and decision-making under pressure. But he also saw companies making critical decisions about security and market entry without understanding what those choices looked like on the ground. Rather than take the conventional route into another government post or a large defence contractor, he wanted to build something independent, rooted in practical experience and commercial discipline.

That logic explains why Al Thuraya sprawls across security, energy, technology, and consulting without looking like diversification for its own sake. The real connective tissue, Padilla explains, is risk. In the markets where his clients operate, a security issue quickly becomes an operational problem, which becomes a financial one, and a technology weakness can just as easily become a security vulnerability.

Humans, he notes, are often the risk themselves, through poor judgement, inaction, or simply misreading an environment. Most of the group’s businesses emerged organically, as one client problem exposed another next door.

Even something as simple as counting the group’s holdings reveals this philosophy. Public materials have cited both 17 and 18 companies, a discrepancy Padilla attributes to a document that wasn’t updated after a new entity was created. But he pushes back against vanity metrics altogether, arguing that wholly owned operating companies, joint ventures, and purpose-built entities don’t all carry equal size or strategic weight. He’d rather be precise about what’s actually functioning than inflate a number for appearances.

Also Read: The founder-to-minister pivot isn’t the problem, ASEAN’s missing governance infrastructure is

Looking ahead, Padilla is explicit that the next chapter isn’t about acquiring more companies. The priority is deepening capability where it already exists, strengthening management, and applying technology more intelligently.

Expansion will be selective, and only where genuine client need justifies it, as seen in his cautious approach to Southeast Asia, where work has followed existing clients rather than a deliberate strategic push. He’s equally willing to sell or restructure businesses if another owner could scale them faster, rejecting the founder’s instinct to cling to every asset out of sentiment.

Leadership forged between two worlds

One of the clearest illustrations of Padilla’s evolution as a leader came early in Al Thuraya’s history, when an employee was kidnapped. His instinct was purely operational: mobilise, identify local influencers, and act decisively. The client, however, saw the crisis through a different lens entirely, one bound by insurance protocols, legal counsel, and HR procedures.

Neither instinct was wrong; they were simply different frameworks for the same emergency. Padilla borrowed a principle from his military past — that the enemy always gets a vote — and applied a civilian corollary: the client gets a vote too.

That recalibration extended to how he manages information. One Special Operations habit he had to consciously abandon was compartmentalising knowledge on a need-to-know basis. In the military, restricting information protects missions; in business, it strangles ownership. Employees who don’t understand why a decision was made can’t be expected to take real responsibility for executing it.

His early international experience taught a parallel lesson: Western assumptions about authority and process don’t automatically translate abroad. Seniority on an org chart doesn’t always equal influence, and in many markets, relationships must be established before transactions can follow.

For Padilla, authentic leadership began only after he left the structured military hierarchy behind. Without rank to command compliance, he found himself personally handling sales, accounting, and project management in Al Thuraya’s early days, a grounding experience that taught him people follow credibility, not titles. Leadership, he says, is the relationship between judgement and responsibility: making a call, explaining it, owning the outcome, and reversing course when proven wrong.

That same unsentimental logic shapes his stance on governance. Padilla has not separated the chairman and CEO roles, a structure he attributes to Al Thuraya still being founder-led, but he’s open to changing it if scale or institutional investment ever demands it. “I did not create the group to protect a title,” he says.

Resilience, risk, and moral boundaries

Padilla is sceptical of companies that call themselves resilient, only to evacuate staff or freeze operations at the first sign of trouble. For him, resilience isn’t a personality trait; it’s an engineered system, built by confronting uncomfortable questions long before a crisis arrives.

What happens if a major client disappears? If banking access or communications suddenly fail? The answers, he argues, aren’t found in motivational speeches but in cash discipline, cross-trained staff, and contingency planning done while decision-makers still have time to think clearly.

Also Read: The cloud is just someone else’s computer. Sometimes that computer gets hit by a drone

The same discipline applies to Al Thuraya’s own exposure. Operating in difficult environments by design means absorbing political, economic, and security risk, but Padilla draws a firm line between advising a client and making their decisions for them. The group’s responsibility is to deliver accurate, honest analysis; the client decides how much risk to accept.

Operating across governments with varying human rights records forces constant ethical calibration. Padilla is unambiguous: Al Thuraya will not knowingly provide services aimed at civilians, unlawful repression, or objectives it cannot defend. Legality, he insists, isn’t the only test. Something can be perfectly legal and still not be work he wants the company’s name attached to. His personal benchmark is simple: could he honestly explain the engagement to his employees and to himself? If not, the answer is no.

Reflecting on where his judgement has failed him, Padilla points not to fundamental misreads of dangerous environments, but to the gap between what his teams observe on the ground and how clients choose to interpret that intelligence. He’s also candid about his own missteps, chiefly letting the attractiveness of an opportunity outpace its commercial fundamentals. Real failure, in his view, isn’t closing a company when conditions shift; it’s continuing to fund one out of stubbornness rather than strategy.

Technology with ground truth attached

Padilla’s approach to artificial intelligence mirrors his broader philosophy: useful when paired with human judgement, dangerous when substituted for it. Al Thuraya deploys AI for open-source intelligence, anomaly detection, due diligence, and forecasting, but always alongside people actually embedded in the markets being analysed (Padilla himself lives in North Africa).

He remains openly sceptical of predictive threat-scoring claims, viewing much of that language as marketing ahead of operational reality. His test for any technology is blunt: does it improve a decision, an outcome, cost, or safety? If not, it’s just another slide in a presentation.

Also Read: “The AI did it” is not a defence; it is a confession

That caution extends to autonomous decision-making. Padilla is comfortable with AI flagging anomalies but uneasy about allowing unreviewed systems to make calls affecting someone’s liberty, safety, or reputation. Speed, he argues, is no justification for removing accountability; a bad process made faster is still a bad process. The more powerful AI becomes, he insists, the more important human responsibility becomes.

The measure of what comes next

Asked what his younger, uniformed self would make of what he’s built, Padilla doesn’t hesitate: surprise at the sheer breadth of sectors and countries involved, and unease at how much still depends on one individual. The question that once drove him “how much can we build?” has been replaced by a more demanding one: is what we’ve built strong enough to continue without me? Building the group, he reflects, was one challenge. Making sure it matures, adapts, and continues responsibly is the next one.

The post Michael Padilla: The costliest failure is refusing to stop when strategy no longer holds appeared first on e27.

Posted on — Leave a comment

Zeya Health acquires ConnectLah to build AI booking rails for clinics

Zeya Health founder and CEO Agastya Samat (L) and ConnectLah  founder Jules Pereira-Gomes

In healthcare, booking an appointment is rarely as simple as finding an empty slot on a calendar. A patient may need a specific doctor, a particular procedure, a referral, a follow-up window, or a clinic that accepts certain rules around timing and availability. Much of this still sits in phone calls, WhatsApp threads and front-desk work.

Singapore-based Zeya Health is betting that this messy layer of healthcare access is where AI can create real value. The Antler-backed startup has acquired ConnectLah, a clinic automation company that helps healthcare providers manage patient communication and booking workflows through WhatsApp.

Also Read: Graas raises US$17M, acquires Trustana to build smarter retail AI agents

The deal brings ConnectLah’s product, technology and customer relationships into Zeya’s patient access platform. Financial terms were not disclosed.

Zeya said the acquisition comes after a year in which it recorded 15-fold revenue growth. The company currently works with healthcare providers in Singapore, Malaysia, Australia, Indonesia and Vietnam, as well as a hospital partner in Cambodia. It aims to reach 1,000 healthcare providers by mid-2027.

From chat automation to AI booking infrastructure

Zeya already helps clinics and hospitals manage patient enquiries, bookings, reminders and follow-ups across messaging and voice channels. With ConnectLah folded into the company, Zeya wants to deepen its ability to automate the steps between a patient asking for an appointment and a provider confirming it.

The company describes this as “agentic booking infrastructure”. In plain terms, that means building systems that allow AI agents to do more than answer questions. They must be able to understand a patient’s request, check clinic rules, match appointment types with the right practitioner, interact with existing healthcare software, and complete a booking without breaking the provider’s workflow.

“Booking an appointment should be as simple as asking for one,” said Agastya Samat, founder and CEO of Zeya Health. “The complexity sits behind the scenes: the schedules, systems and rules that determine what can actually be booked.”

That complexity is particularly visible in Southeast Asia, where healthcare systems are highly fragmented. Large private hospital groups, specialist centres, neighbourhood clinics and independent practitioners often use different software systems, communication habits and administrative processes. Many still rely heavily on WhatsApp, calls and manual coordination, especially in markets where consumer messaging apps have become the default front door for services.

Also Read: The future of healthcare AI isn’t more data. It’s better context

This makes healthcare access a difficult problem for generic AI tools. A chatbot can collect a patient’s name and preferred time, but it cannot safely confirm an appointment unless it understands the provider’s rules and can connect to the relevant system of record. Zeya’s argument is that the real opportunity lies not in another patient-facing bot, but in the infrastructure layer that lets AI work reliably with healthcare operations.

Why ConnectLah matters

ConnectLah’s focus on WhatsApp gives Zeya a practical entry point into how many clinics already communicate with patients. The company helps automate patient communication, appointment booking, reminders and front-desk workflows inside familiar messaging channels.

For Zeya, the acquisition is less about adding a standalone product and more about absorbing workflows that have already been tested in clinics. ConnectLah customers are expected to gain access to a broader set of tools connected to existing clinic systems, while Zeya expands its provider footprint.

“ConnectLah brought AI into the channels clinics and patients already use, proving the approach in real clinics and live patient workflows,” said Jules Pereira-Gomes, founder of ConnectLah, who will join Zeya as a strategic advisor.

The transition will focus on the workflows and integrations relevant to each provider, according to the company. In practice, that means Zeya will need to account for variations in how clinics handle different appointment types, practitioner availability, reminders and follow-ups.

If executed well, the combined platform could reduce the back-and-forth that often defines healthcare booking. A patient asks a question, a staff member checks availability, clarifies details, proposes a time, waits for confirmation, then sends reminders manually. For providers facing staff shortages or rising patient volumes, shaving down this coordination burden can be more than a convenience.

A regional race for the healthcare front door

Zeya is entering a crowded but still unsettled market. Across Asia Pacific, startups and incumbents are trying to own different parts of the healthcare access stack.

At the consumer-facing end, companies such as Doctor Anywhere in Singapore, Halodoc in Indonesia, Practo in India and HealthEngine in Australia have built large patient networks around online consultations, appointment discovery or health services marketplaces. Globally, firms such as Doctolib in Europe and Zocdoc in the US have shown how appointment booking can become a major platform category.

Also Read: Vietnam’s healthtech boom has a talent problem nobody is talking about

Zeya’s positioning is different. Rather than building primarily as a consumer marketplace, it is targeting the provider infrastructure layer: the workflows behind bookings, reminders and patient communication. That could make it complementary to some patient-facing platforms, but it also places the company in competition with clinic management systems, patient engagement tools and AI automation vendors trying to modernise healthcare administration.

The distinction matters in Southeast Asia. Many healthcare providers are cautious about surrendering patient relationships to third-party marketplaces. A tool that helps clinics keep control of their booking rules and communication channels may face less resistance than one that tries to redirect patients into a new app.

The AI assistant angle

Zeya’s longer-term ambition is to make healthcare booking workflows accessible to authorised personal AI assistants. The idea is that a patient could ask their preferred AI assistant to find and book an appointment, while the provider’s systems still enforce the rules around what can actually be scheduled.

This is still an emerging behaviour. Most patients today are not using personal AI agents to book medical appointments. But the direction is clear: as AI assistants become more capable, they will need reliable rails into real-world services. Healthcare is one of the hardest categories because accuracy, privacy, consent and operational fit all matter.

That is why Zeya’s bet is both timely and difficult. The company is not just trying to automate clinic chat. It is trying to become part of the connective tissue between patients, providers and the AI agents that may increasingly sit between them.

Also Read: The app worked, the product didn’t: Can we install judgement into AI agents?

For now, the acquisition of ConnectLah gives Zeya more product depth, more customer relationships and a faster route into clinics already using messaging-led workflows. The bigger test will be whether it can scale across countries where healthcare software adoption, regulation and clinic behaviour vary widely.

In a region where the first point of care is often still a phone call or a WhatsApp message, the company’s opportunity is clear: make booking feel simple for patients without pretending the backend is simple for providers.

The post Zeya Health acquires ConnectLah to build AI booking rails for clinics appeared first on e27.

Posted on — Leave a comment

Uptober or downtober: Will Bitcoin’s 19% seasonal average survive US$100 oil?

The question on every trader’s mind as October begins is whether Bitcoin will deliver its legendary Uptober performance or succumb to the economic pressures that have defined much of this year. The answer, based on the data, is neither a triumphant rally nor a catastrophic collapse. It is something more nuanced and arguably more challenging: a month of range-bound volatility that rewards discipline over conviction.

Bitcoin trades at approximately US$83,070, wedged tightly between crucial technical structures and economic pressure points. This price level is not random. It reflects a market that absorbed significant leverage flush late last month and now depends heavily on spot order books for direction. The immediate trend hangs in the balance, and the battle between seasonality and economic headwinds has left crypto markets heavily divided as Q4 begins.

The bull case rests on a foundation of historical precedent and technical momentum. Between 2013 and 2025, Bitcoin averaged around 19 per cent gains in October and closed the month in the green 10 out of 13 times. This track record earned Uptober its reputation as a psychologically powerful sentiment driver. The momentum setup supports this narrative. BTC has logged consecutive monthly gains heading into Q4. If the market reclaims and firmly holds above US$84,000 to US$84,433, a technical path opens toward major resistance at US$87,360 and psychological levels near US$90,000.

The altcoin rotation signal adds another layer to the bull thesis. The OTHERS/BTC chart is testing resistance, and an expected rollover in Bitcoin dominance points to early liquidity rotation into majors like Ethereum, which saw US$624.1M in weekly ETF inflows. This suggests capital is beginning to explore beyond Bitcoin, a classic precursor to broader market strength.

Also Read: The sovereign shift: Why nation states are trading gold for Bitcoin

The bear case is equally compelling and grounded in present realities rather than historical patterns. October is never a guaranteed win. Last year, geopolitical and tariff threats drove a massive US$19 billion liquidation event that wiped out the Uptober narrative, leaving the month down roughly 4 per cent. That episode reminds traders that exogenous shocks can override seasonality.

The present economic picture offers several such shocks in waiting. Brent crude holds above US$100 per barrel due to ongoing conflict around the Strait of Hormuz. Energy-driven inflation is a lingering risk that feeds directly into consumer prices. US CPI sits at 3.4 per cent YoY, keeping fixed-income yields highly competitive. The 10-year Treasury yield hovers above 5 per cent, creating an explicit hurdle for risk assets. Ahead of the pivotal October 27 to 28 FOMC meeting, the market is bracing for another potential interest rate hike. These are not abstract concerns. They are concrete headwinds that constrain the upside for Bitcoin and other risk assets.

The synthesis of these opposing forces leads to a clear conclusion. Unless institutional ETF inflows dramatically surge past US$1 billion daily, the combined pressure of expensive oil, high yields, and monetary tightening will likely confine Bitcoin to a defined trading channel. A straightforward replication of the historical 19 per cent October gain is highly challenging in this environment. Instead, expect a highly volatile start to the month with major support anchoring near US$80,811 and deeper liquidity pools resting around US$74,000 to US$75,585 if economic conditions deteriorate further.

This outlook has direct implications for how participants should operate. The split between short-term leverage trading and spot positioning for Q4 requires entirely different operational frameworks given the current economic landscape.

For leverage traders, the arena is less susceptible to cascading 10 per cent flash crashes because futures open interest has levelled out around US$53 billion. It remains highly prone to stop hunting. Major options max-pain levels sit below the current price. If Bitcoin attempts to rally but repeatedly fails to break the US$85,000 resistance barrier, scaling into short positions targeting an inefficiency sweep back toward US$80,875 becomes a viable strategy. Do not chase longs inside the current cluster. Wait for a definitive daily close above US$85,000. Reclaiming this level triggers a short-squeeze vector toward US$87,397, with a final target near the US$90,000 psychological barrier.

Also Read: Can Bitcoin hold US$82,000? Inside the security fear and macro storm

For those positioning for the entirety of Q4, the entry strategy should anticipate economic friction in late October. With the 10-year Treasury yielding 5.17 per cent and oil above US$100, the market will likely experience a mid-month liquidity drain. Treat any geopolitical or economically driven pullbacks into the US$74,000 to US$75,585 demand zone as a high-probability spot-buy tier.

On the altcoin front, Bitcoin dominance remains elevated at 58.67 per cent. Capital is not yet flowing freely into high-beta assets. Keep spot allocations concentrated heavily in large-cap majors like Ethereum or Solana until Bitcoin dominance drops cleanly below 58 per cent, which will act as the green light for broader altcoin exposure.

My perspective is that the Uptober narrative, while emotionally satisfying, distracts from the structural reality. The market is not in a phase where historical averages dictate outcomes. It is in a phase where economic conditions set the boundaries, and technical levels define the trading range.

The most successful participants this month will be those who respect the range, manage risk around the FOMC meeting, and position for Q4 through patience rather than fear of missing out. The battle between Uptober and Downtober will not produce a winner in the traditional sense. It will produce a grinding, volatile month that rewards those who understand the difference between a seasonal pattern and a structural trend.

—

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

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

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

The post Uptober or downtober: Will Bitcoin’s 19% seasonal average survive US$100 oil? appeared first on e27.

Posted on — Leave a comment

Why Singapore firms fear data sovereignty failures but remain underprepared

Singapore’s position as one of Asia’s most advanced digital economies is built on a simple promise: global companies can move data, capital and operations through the city-state with confidence. A new study suggests that promise is becoming harder to keep.

Research released by data storage and management company Everpure found that 89 per cent of Singapore-based enterprise leaders believe a data sovereignty failure could cost them their jobs. The fear is not only personal. The same proportion said such a failure could damage their organisation financially and reputationally.

Also Read: The new border: Why server farms are the battleground of AI sovereignty

Yet the more striking finding is the gap between concern and action. According to Everpure’s Global Data Sovereignty Report 2026, 81 per cent of Singaporean enterprises surveyed do not have a formal data sovereignty strategy in place, the highest share among the eight markets covered in the study.

That matters because data sovereignty is no longer just a legal question about where information is stored. It has become a business continuity, geopolitical and vendor risk issue.

At its simplest, data sovereignty refers to the idea that data is subject to the laws and controls of the country or jurisdiction in which it is stored, processed or accessed. In practice, the challenge is more complicated: companies must know who can access their data, which foreign laws may apply, and whether a cloud or software provider could be forced to hand over information or suspend services during a political dispute.

For Singapore, a regional headquarters for banks, tech companies, logistics players and digital platforms, the issue cuts especially close. The country’s economy depends on trusted cross-border flows of information. At the same time, its companies often rely on global cloud, software-as-a-service (SaaS) and cybersecurity vendors whose infrastructure may span several jurisdictions.

Awareness is high, preparedness is not

Everpure commissioned research firm Vanson Bourne to survey 2,100 C-suite and IT leaders from large enterprises across the UK, France, Germany, Australia, Japan, South Korea, Singapore and India in June 2026. Singapore accounted for 100 respondents.

The study found that 93 per cent of Singapore organisations recognise data sovereignty as a business concern, compared with 90 per cent globally. Some are already changing procurement behaviour. About 41 per cent of Singapore respondents said they are limiting their use of SaaS providers that rely on non-domestic infrastructure, while 86 per cent said they would compromise on advanced features to work with a local or sovereign provider.

Also Read: Should cybersecurity be nationalised?

But recognition has not translated into operational readiness. In Singapore, 63 per cent of enterprises said they lack full visibility into who can access, control and manage their data. More worrying, 68 per cent said they have no mitigation plans for geopolitical data exfiltration or service disruption.

This is the heart of the “sovereignty gap” highlighted in the report: executives know the risk is material, but many organisations have not built the governance, technical controls or response plans needed to manage it.

Nathan Hall, Vice President and General Manager for Asia Pacific and Japan at Everpure, said the risk for Singapore lies in the disconnect between digital maturity and organisational preparedness.

“Singapore is one of the most digitally mature markets in the world, yet 81 per cent of enterprises here are operating without a formal data sovereignty strategy. That gap between awareness and action is the real risk,” he said. “Sovereignty is not simply about where data sits — it is about knowing who can access it, which jurisdictions apply, and whether critical services could be disrupted.”

The point is especially relevant in Southeast Asia, where regulation is still uneven across markets. Singapore has a mature data protection regime under the Personal Data Protection Act, while neighbouring economies are developing or refining their own privacy, cybersecurity and localisation rules. For regional companies, this creates a patchwork problem: data may be generated in Indonesia, processed in Singapore, analysed through a US-headquartered SaaS platform, and stored on infrastructure distributed across several markets.

The AI factor

The sovereignty question is becoming more urgent because of artificial intelligence. As companies feed more enterprise data into AI systems, the boundaries around storage, access and reuse become harder to track. Sensitive operational data may move into model-training environments, analytics platforms or third-party applications without executives fully understanding where it goes or how it is governed.

Also Read: AI governance is moving from promises to proof

This is not just a theoretical risk. Banks, insurers, healthcare groups and government-linked enterprises in Southeast Asia are under growing pressure to adopt AI while maintaining strict controls over customer data. For startups and scaleups, the challenge is different but no less serious. Many depend on global cloud platforms and AI tools from day one, often without the resources to conduct deep vendor risk reviews.

Everpure’s survey suggests that companies are still treating sovereignty as an extension of cybersecurity or compliance. That may be too narrow. Cybersecurity focuses on preventing unauthorised access or attacks. Compliance focuses on meeting legal obligations. Sovereignty adds another layer: whether an organisation retains effective control over its data when foreign laws, vendor dependencies or geopolitical shocks come into play.

Rahiel Nasir, Research Director and Lead Analyst for Worldwide Digital Sovereignty at IDC, described the shift as a board-level issue. “The challenge for executives is not just about knowing where their data are hosted,” he said. “It is about being in total control of all data access and transfers, including all metadata, guaranteed protection against extra-territorial data requests, and managing IT and vendor risks in the light of geopolitical uncertainties.”

From compliance checklist to operating model

Everpure argues that companies should move towards “sovereignty by design”, where governance and controls are applied based on the risk of each data set, application and workload. In practical terms, that means mapping critical data, classifying it properly, understanding vendor access, and deciding which workloads require stricter controls.

The distinction matters. Not every piece of enterprise data needs the same level of protection. A marketing dashboard, payroll file and national infrastructure system carry different risks. A blanket localisation strategy can be expensive and restrictive; a purely global cloud approach can leave companies exposed. The harder but more useful path is to decide which data must remain under tighter corporate control and which can safely sit within global platforms.

Also Read: Southeast Asia’s AI buildout is racing toward a power wall

For Singapore, the findings should be read less as an indictment and more as an early warning. The country has spent years building itself into a trusted digital hub for Asia. Maintaining that position will require not only strong national regulation, but also stronger internal discipline among enterprises using cloud, SaaS and AI systems.

The boardroom fear captured in Everpure’s report may sound dramatic. But in a region where data flows underpin finance, trade, healthcare and digital services, sovereignty failures are no longer abstract policy debates. They are operational risks, and increasingly, leadership risks.

The post Why Singapore firms fear data sovereignty failures but remain underprepared appeared first on e27.

Posted on — Leave a comment

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

Indonesia’s agritech sector looks different in 2026 than it did when TaniHub raised its Series B in 2021. The cohort of platforms that emerged in the late 2010s — TaniHub, Crowde, iGrow, Sayurbox, and others — aimed to do for smallholder agriculture what fintech had done for SME credit: build technology infrastructure for a segment the formal banking system could not serve at scale.

Some platforms have grown. Several have restructured significantly. A few have wound down. The credit arm of the sector, in particular, has been through a harder cycle than founders or investors anticipated.

After fifteen years inside Indonesian risk functions, I have watched this cycle with direct visibility into the structural risk questions it raised. The story underneath the platform-level outcomes is more useful than the company-by-company narrative.

What the first agritech generation built

The Indonesian agritech sector that emerged after 2016 built two adjacent infrastructures. On the commerce side, platforms built marketplaces connecting smallholder farmers directly to institutional and retail buyers — restaurants, hotels, supermarkets, modern trade. On the credit side, P2P lending platforms targeted agricultural working capital specifically — seeds, fertiliser, equipment, harvest financing.

TaniHub, founded by Pamitra Wineka, Ivan Arie Sustiawan, William Setiawan, and Michael Jovan in 2016, was one of the most prominent platforms to combine both sides through its TaniFund credit arm. Its Series B in 2021 was one of the largest agritech rounds Indonesia had seen.

Also Read: Why Southeast Asian agritech must build for acquisitions, not IPOs

What turned out to be harder than expected

Three things proved structurally more difficult than early models priced in.

  • Harvest cycle credit timing. Agricultural loans do not behave like consumer or SME credit. They are paid back from a single harvest event that can be three, six, or twelve months out, and that may or may not arrive at the expected volume or price. A platform’s cash flow timing assumes regular monthly repayments. A farmer’s cash flow timing does not.
  • Default correlation. Agri-credit defaults are not independent in the way SME or consumer defaults mostly are. A drought, a flood, a pest event, a commodity price collapse — any of these correlate defaults across hundreds of borrowers in the same region simultaneously. Models that treat each loan as independently distributed misprice this correlation, sometimes by an order of magnitude.
  • Operational cost per loan. The cost of underwriting, monitoring, and collecting a small agricultural loan — often spread across remote geographies, often involving in-person verification — is high relative to the loan ticket. Platforms that priced loans against urban operational assumptions ended up subsidising rural credit from other revenue lines.

Lessons learned

Six principles emerge from the Indonesian agritech credit cycle.

  • Agri-credit is not consumer credit with mud. The underlying risk physics — harvest cycles, weather correlation, commodity volatility — is structurally different. Models built for one will misprice the other.
  • Correlation is the silent killer. The single biggest pricing error in early agritech credit was underestimating how correlated defaults can become inside a single weather or price event. Diversification across crops, geographies, and harvest cycles is solvency infrastructure, not a marketing point.
  • Operational cost is destiny. Platforms that did not build for the cost of remote, small-ticket lending from day one ended up cross-subsidising it forever — or stopped lending. There is no version of agritech credit at scale without operational cost discipline designed for the segment.
  • Funding tenor must match harvest tenor. Short-term retail or institutional funding does not pair well with agricultural cash flows. The platforms that survived better had funding partners willing to hold positions across full agricultural cycles, not calendar quarters.
  • The buyer side is more durable than the credit side. The commerce infrastructure built by Indonesian agritech — connecting farmers to buyers, aggregating produce, building cold chain — has aged better than the credit infrastructure built alongside it. Future capital allocation should reflect that.
  • The credit gap is durable, the model needs to mature. Indonesian smallholder agriculture will still need credit access whatever the platform-level outcomes of the first generation. The next models will need to absorb the lessons the first generation paid for, or they will pay for them again.

Also Read: Agritech investors are learning that infrastructure matters

The macro stakes

TaniHub and the agritech cohort built something Indonesia had not had before: a market-facing technology layer on top of smallholder agriculture. Not all of it survived. The parts that did, and the lessons from the parts that did not, are now the foundation for whatever comes next.

Indonesian smallholder agriculture employs tens of millions of people and produces a significant share of the country’s food supply. The credit gap at the farm gate is one of the most important development finance questions in Southeast Asia. The first generation showed that the gap could be addressed by technology platforms — and that it is harder than the original models predicted. The next generation has the data to do better. The opportunity is still there.

—

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

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

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

The post The agritech credit paradox: Lessons from TaniHub and Indonesia’s first agritech generation appeared first on e27.