SCBX’s 3-Layer AI Roadmap: A Blueprint for Scaling AI in Big Companies

As more companies start to grasp AI's potential, the question is no longer just "how do we experiment with AI?" It's how to move beyond experimentation entirely by integrating AI into the core structure of the organization to create real economic impact.

In the Techsauce Global Summit session "AI-First in Action: Inside an Enterprise AI Transformation," Techsauce CEO Oranuch Lerdsuwankij sat down with Kaweewut Temphuwapath, Chief Innovation Officer of SCBX and CEO of SCB 10X, to discuss the direction SCBX, a technology group worth more than 20 billion USD, is taking across its three generations of business:

• Traditional banking (Gen 1)
• Digital lending (Gen 2)
• Digital assets and SCB 10X, which manages over 600 million USD in venture capital, driving the group's push to become a genuinely AI-first organization (Gen 3)

This is the summary of the session, distilled into a practical framework for any leader trying to transform both themselves and their company.

Defining "AI-First": Start With an Ambitious Goal, Not a Definition

"AI-first" or "AI-native" risks becoming just another buzzword, so we must undefine it first. SCBX chose not to get bogged down in technical definitions, and instead set bold, quantifiable goals nearly two years ago to rally the organization around a shared direction:

• 75% AI-enabled revenue: a target for 75% of income to be generated with AI's direct involvement
• 15% AI talent: a goal for 15% of the workforce to be classified as genuine AI talent
• Adoption speed matched to the business: recognizing that different parts of the organization can't move at the same pace. The traditional bank (Gen 1) has to prioritize system stability and risk controls, while Gen 2 and Gen 3 businesses can move much faster in adopting new technology

A Three-Layer Architecture to Escape the "Pilot Trap"

One of the biggest challenges for large organizations is getting stuck in a sandbox structure where dozens of scattered AI initiatives never translate into actual profit. To avoid this, SCBX organized its AI strategy into three layers, giving the entire organization a shared framework:

• Layer 1 — Use Cases (focused on P&L impact): Every flagship project must have a clearly measurable effect on profit and loss, whether that means increasing revenue or cutting costs. Critically, these must be business-driven initiatives, not IT projects.
• Layer 2 — Capabilities (focused on reusability): Build "central capabilities" that any team across the organization can plug into. A Thai-language conversational AI, for example, only needs to be built once then it can be adapted for sales, customer service, or collections across the business.
• Layer 3 — Infrastructure & People (focused on scalability and governance):
  • Agentic platform: a foundational structure that allows AI agents to work together precisely, with tight control over latency and cost
  • Enabling processes: reworking compliance and risk management so they accelerate AI adoption rather than becoming a bottleneck
  • Single backlog: consolidating and prioritizing every AI project across the company into one unified backlog, simplifying investment decisions

The Typhoon Case Study and Thailand's Sovereign AI Strategy

Two years ago, when SCBX first began developing Typhoon—its own Thai-language large language model—the effort became a defining lesson for the company's technology leadership. Rather than starting from scratch, the team built on top of leading open-source models like Llama and Mistral, fine-tuning them to achieve a breakthrough in Thai-language performance that other models hadn't yet matched.

Over time, as top global models improved their own Thai-language capabilities, SCBX shifted strategy by adopting a mix of different open-source models rather than relying on a single one. But the most valuable asset to come out of the process wasn't the model itself. It was the talent pool: the team of engineers with deep expertise in fine-tuning and token optimization.

In business terms, the real prize is cost advantage where the organizations that win will be the ones that drive down the cost per token while holding quality steady. Today, Typhoon has evolved beyond being just a model the company built; it now serves as a hub for spreading AI knowledge across the organization, helping every department apply the right technology effectively.

Building an AI Culture at Scale

The heart of becoming an AI-first company isn't having the most advanced technology — it's motivating people and building an AI culture across the entire enterprise. That means bringing every team and every generation of employee together.

Shift the mindset: AI is the new baseline skill

SCBX draws a compelling comparison: in the near future, AI proficiency won't be optional anymore, it will be as fundamental as typing, using email, or knowing Excel and PowerPoint. Today, a candidate who said "I don't know how to use email" or "I can't use a computer" would likely be screened out immediately. AI skills are heading toward that same standard. Every employee will need to adapt in order to stay valuable and keep pace with the technology defining this era.

Attitude matters more than skill

From a leadership perspective, teaching AI skills isn't the difficult part, there's no shortage of courses and training available. The real challenge is attitude. A common pattern: an employee tries AI once, decides "it's not as good as a human," and gives up entirely. In reality, this often just means they haven't yet learned how to prompt effectively or experimented enough. Building the right AI culture means encouraging open-mindedness and persistence rather than abandoning the effort after one try.

Turning training into a lasting community

To drive this shift, SCBX trained 100% of its employees on Copilot over the past year. But classroom training alone isn't enough to build a lasting culture. Therefore SCBX also built an internal community centered on three things:

• Best practices: identifying employees who've used AI to solve real pain points in their day-to-day work
• Democratizing knowledge: having those employees share their techniques and experience with their peers
• Building momentum from within: seeing a colleague get faster, better results with AI naturally inspires others to learn, but making that happen at scale requires management and HR to actively gather these stories and spread them across the organization

SCBX's transformation into an AI-first company shows that success doesn't come from hoarding the most advanced technology or running hundreds of lab experiments. It comes from a clear vision, a functional system, and bringing people together to move forward as one team.

Source: Session on ‘AI-First in Action: Inside an Enterprise AI Transformation’ hosted by Ms. Oranuch Lerdsuwankij, CEO of Techsauce, and Mr. Kaweewut Temphuwapath, Chief Innovation Officer of SCBX and CEO of SCB 10X at Techsauce Global Summit 2026