From Sugar Cane to AI Agents: The Foundational Secret to Thailand's Data Revolution

The truth about AI adoption isn’t complexity — it’s that most companies are building skyscrapers on shaky foundations. While executives chase the latest AI buzzwords, they're sometimes overlooking the reality that determines success or failure: data infrastructure.

At a recent IBM Executive Summit in Bangkok, this reality crystallized through Athikom Kanchanavibhu, executive vice president of digital and technology transformation at Mitr Phol Group — Thailand’s sugar, energy, and bio-based products giant. He didn’t just talk about AI adoption; he shared insights learned from transforming a traditional agricultural conglomerate into an AI-driven enterprise across a vast network of subsidiaries across nine countries.

“Don’t try to dump everything into the data lake,” Kanchanavibhu warned during the event’s panel discussion. “People say data is our asset, but actually, it becomes a liability if you just throw everything in. We need to start from the business case. What are the business values that we need to drive from curated data and AI?”

This isn’t just one company’s learning curve but a preview of Thailand’s AI future. As the country positions itself as a regional AI hub, early adopters like Mitr Phol are helping to write the playbook for what comes next.

The data paradox: Why more isn’t always better

Nikhil Singh Kushwaha, IBM’s chief technology officer for data and AI in Asia Pacific, delivered a key insight at the start of the event, which was entitled “Accelerate Growth with an AI Ready Data Architecture”: “If you keep feeding garbage into a model, the insight or output will never be accurate. More data leads to better AI is actually a myth.”

The problem isn’t data scarcity but the need for better curation. Thai companies are generating significant amounts of data from IoT sensors, customer interactions, financial systems, and operational processes. But most may lack the fundamental infrastructure to transform this digital exhaust into AI fuel.

IBM’s global research confirms this challenge. Brian Goehring, associate partner and AI research lead at IBM’s Institute for Business Value, revealed that Thai organizations are predominantly in “developing or emerging stages” of data maturity. He cited an IBV benchmarking study of 2,500 organizations globally that showed persistent data silos as a critical barrier.

Mitr Phol’s journey illustrates this evolution. The company spent years building a data lakehouse architecture, integrating everything from farm sensors to GPS tracking to traditional ERP systems. But the paradigm shifted when they stopped trying to capture everything and started asking: What business problems are we solving?

“We went back and talked to the business units from upstream to downstream and asked what the business cases were,” Mitr Phol’s Kanchanavibhu explained. “And we became more selective about what data we put into the lake and what AI use cases we built.”

This selectivity is less about costs and more about clarity. When humans struggle to understand their own data, how can machine learning models generate meaningful insights?

The ROI reality check

Anothai Wettayakorn, IBM Thailand’s managing director, offers a practical perspective on AI with hard numbers. IBM’s own AI transformation delivered USD3.5 billion in productivity gains by 2024, exceeding projections by 75%. But the key insight isn’t the number; it’s the measurement framework.

Goehring’s research tracks a crucial evolution: “Following the hype of generative AI, when we saw the return on investment being expected to be in the sort of teens, 20s and even 30% — well above the cost of capital — that’s sort of unbelievable. We interpreted that to mean that it was really based on pilots, not necessarily scalable production.”

Those inflated expectations have normalized to “high single digits of ROI,” but something more significant is happening. Companies are “shifting their mindset from a very project-based view… to what’s happening to the bottom line.”

Thai companies are moving beyond the experimental phase and are decisively shifting into active GenAI implementations. As Vatsun Thirapatarapong, AWS Thailand’s country manager, observed, “The experimental phase is behind us. Organizations are now focused on strategic deployment and targeted value extraction from AI implementations. Looking ahead, AI is becoming a fundamental driver of business transformation, with Thai enterprises actively exploring how to leverage AI agents, transform entire operations, and foster organization-wide AI adoption. We're seeing AI being deeply embedded into workflows, scaling across enterprises, and unlocking unprecedented levels of efficiency and innovation. The key questions leaders are asking have evolved from 'Should we adopt AI?' to 'How can we maximize AI's impact across our entire organization?”

The new framework extends beyond traditional ROI to include Return on Employee (productivity gains) and Return on Future (strategic positioning). For Thai enterprises, this means AI investments must demonstrate immediate operational improvements while building capabilities for emerging opportunities.

From pilots to production: The governance gap

Here’s where most Thai companies often face challenges: they successfully nail the proof-of-concept but may struggle to scale it. Mitr Phol has developed numerous AI models across operations, with seven now live in production. But getting to that point required solving challenges rarely addressed in AI vendor presentations.

“We reviewed the data warehouse and integrated data from farm to table — from sensor images and GPS signals to structured data from ERP, CRM, and customer systems — all into our data lakehouse,” Mitr Phol’s Kanchanavibhu shared. “From there, we began developing AI solutions.”

The governance aspect is crucial beyond individual companies. Goehring’s research reveals that Thai organizations lag behind their global peers in generative AI adoption, with governance maturity presenting both challenges and opportunities. “Your business case can be solid, your strategy can be fantastic, and the model design can be phenomenal. But if there’s a quality issue [with the data] and the people who need to engage with it don’t trust it… the business case doesn’t close.”

Mitr Phol’s architecture is built upon three foundational layers: data — comprising a lakehouse integrated with a data fabric to unify heterogeneous sources; AI — leveraging composable building blocks to support a variety of models; and interface — an API-driven layer designed with agent orchestration capabilities for seamless interoperability.

But architecture alone isn’t enough. IBM’s Singh highlighted the governance consideration: “Most of the organizations have no idea what is the golden source of their data, whether the data is trusted, who owns this data within the organization, because there are multiple silos around the organization.”

The consequences can be significant. One financial services client required 20 people to work for 30 days solely to trace regulatory reporting data lineage — a task that should have taken minutes with proper governance.

Trust becomes the make-or-break factor. As Goehring emphasized, governance isn’t just regulatory compliance: “It’s really helping those who are going to use the output, and whether employees, customers and suppliers can actually trust the data.”

The Agentic AI frontier

While most Thai companies are still figuring out basic AI implementation, forward-thinking organizations like Mitr Phol are preparing for the next evolution: agentic AI systems that can autonomously interact with multiple business processes.

“Now we’re trying to explore agents from Salesforce, SAP Joule, and others,” Mitr Phol’s Kanchanavibhu revealed. “So we need an architecture that can enable agents from different platforms to communicate with each other.”

This represents a fundamental shift from AI as a tool to AI as a collaborator. But it demands an even more sophisticated data infrastructure. As AWS’s Thirapatarapong noted, “This transformation requires robust, scalable cloud infrastructure. Through advanced cloud technologies like Amazon Bedrock, we're enabling intelligent AI agents to collaborate like human teams, each bringing specialized expertise to complex tasks. It's not just about having AI; it's about having the right infrastructure to support it. This foundation allows AI to truly augment human capabilities and drive innovation."

The timing aligns with broader industry shifts toward agentic AI. Goehring previewed IBM’s upcoming research: “We’ll be coming out with a white paper specifically at the AI summit in London in mid-June that’s looking at what clients have been telling us and organizations have been telling us about agentic AI and their adoption there.”

For Thai companies, today’s data architecture decisions will determine tomorrow’s AI capabilities. Organizations building robust, governed data foundations now will be the ones capable of deploying sophisticated AI agents later.

The Thailand advantage: Sovereignty meets scale

Thailand’s emergence as a regional AI hub isn’t accidental. Recent regulatory developments requiring critical infrastructure data to remain in-country have accelerated local cloud adoption. The AWS Asia Pacific (Thailand) Region, which is “sovereign by design,” addresses these requirements while providing the scale needed for AI workloads.

“The Region will offer an in-country infrastructure for Thailand customers with data residency requirements and allow them to run workloads and securely store data in Thailand while serving end users with even lower latency,” AWS’s Thirapatarapong explained. “Thai enterprises can now confidently build advanced AI capabilities while maintaining complete data sovereignty within Thailand's borders. This robust in-country infrastructure ensures both regulatory compliance and exceptional performance for mission-critical workloads.”

IBM is no stranger to the nuances and challenges of data sovereignty. The company has been advocating for the concept of a sovereign cloud, encompassing data sovereignty, operational sovereignty, and digital sovereignty. This concept is now becoming a bedrock for companies looking to balance AI-driven innovation with local and regional compliance.

Despite governance challenges, Thai organizations show strength in data security maturity, which is a crucial foundation for sovereign AI development. This positions Thailand well for building trusted AI systems that meet both regulatory requirements and business needs.

The partnership imperative

Perhaps the most crucial insight from Mitr Phol’s journey is that AI transformation isn’t just a vendor selection exercise but a purposeful ecosystem play. The company leverages both IBM’s data analytics tools and AWS’s cloud infrastructure, with different technologies serving different purposes.

“I don’t think that any technology vendor can do everything on their own,” IBM’s Wettayakorn acknowledged. “We believe that the partner ecosystem is extremely important, not just for ourselves, but for our customers.”

For Thai companies, this means moving beyond single-vendor strategies to orchestrated partnerships that address specific business needs.

The foundation determines the future

The various success stories demonstrate that AI success is less about choosing the right large language model and more about building the proper foundation. Companies that invest in data governance, quality, and architecture today will capture the AI opportunities of tomorrow. Those that don’t may find themselves perpetually stuck in the pilot phase, burning budgets on experiments that never reach production.

Goehring’s research confirms what practitioners like Mitr Phol have learned: limited self-service analytics capabilities, fragmented data management architecture, and inadequate model standardization create persistent barriers to AI adoption. But, these challenges also represent opportunities for organizations willing to invest in foundational capabilities.

As Mitr Phol’s experience demonstrates, the path to AI leadership runs through data infrastructure. For Thailand’s digital transformation, the foundation is everything. The question isn’t whether AI will reshape Thai business but whether Thai companies will be ready when it does.

Image credit: iStockphoto/Puttachat Kumkrong