AI begins to move from experimentation to large-scale deployment

"We are using AI" has become a familiar answer for many businesses in the past two years. In fact, AI appears in many business tasks, such as customer care, programming support, data analysis, word processing or content creation, giving the feeling that technology is quickly creeping into all production and business activities. However, the gap between investment and actual operating capacity to produce results is still large.

According to a survey on enterprise AI roadmaps conducted by Forrester Consulting and FPT, global businesses are shifting their focus from piloting single AI tools to building AI platforms that can operate on a large scale, especially when AI agents are increasingly deeply involved in business processes. A survey of nearly 400 global business and technology leaders found that 51% spend at least 5% of their information technology budget on AI. 26% rated themselves as having a high level of maturity in AI operations, while 56% were still at an average level.

At the announcement ceremony of the strategic cooperation between Microsoft and FPT to promote AI applications in Asia at the end of June, the gap between "AI applications" and "AI transformation" was one of the key topics mentioned. According to Mr. Mayank Wadhwa, President of Microsoft ASEAN, a concept that is currently receiving attention is the Frontier Firm. These are organizations that go beyond the testing phase to deploy AI at scale, instead of stopping at single initiatives.

According to Mr. Pham Minh Tuan, Deputy General Director of FPT, with experience deploying AI at the corporation and its customers, there is still a gap between "used AI" and "AI creates value". He said many businesses invest significantly, even using up the entire year's AI budget in just a few months, but are still "stuck" in pilot projects, not bringing about changes throughout the organization.

AI application is a business model transformation

At the end of 2022, when ChatGPT kicks off the global AI wave, many businesses consider deploying chatbots, AI assistants or programming support tools as the first step of AI transformation. To date, a series of other AI platforms have been put into application, but reality shows that owning an AI tool is not synonymous with creating business value.

Explaining this issue, Mr. Mayank Wadhwa said that one of the common problems is considering AI simply as a technology project. At that time, businesses add AI to old processes without changing operations, management models and work organization. "A challenge many businesses face is focusing on deploying a single AI tool or technology solution, instead of redesigning work processes, training employees, and building models that can operate at scale," he said.

According to a Microsoft representative, AI is essentially a business model transformation. Therefore, for AI to deliver real results, businesses need three important foundations. The first is to change thinking, considering AI as a driving force in rebuilding operations instead of just a support tool. The second is to build a data, security and governance foundation that is trustworthy enough for AI to operate safely. Finally, there is implementation capacity, which helps take AI from ideas into specific business results.

"This is not a unique challenge for Vietnam but a global problem," Mr. Mayank said. According to him, Vietnam has an advantage when workers and businesses access AI faster than many markets in the region, but that advantage will only be promoted if AI is deployed systematically.

Microsoft's Global AI Diffusion Report released in June showed that Vietnam is in the leading group in ASEAN in terms of AI popularization rate, with the application level by the end of the first quarter of 2026 reaching 26.5%, up from 23.5% in 2025. AI is also no longer a technology reserved for technology businesses, but is spreading to many fields such as finance, manufacturing, retail or services.

Meanwhile, from the actual implementation process in Vietnam and many international markets, Mr. Pham Minh Tuan, representative of FPT, said that many businesses have invested heavily in AI but have not been able to expand.

"The industry challenge is no longer whether to invest in AI, but how to scale AI from pilot projects to actual deployment at production and operational scale," he said.

According to him, AI is creating a new workforce of AI agents that can perform many tasks instead of humans. Unlike chatbots that only respond to requests, AI agents have the ability to plan, coordinate many tools and perform a relatively complete chain of work.

This opens up opportunities to automate more complex processes, but also places new demands on data, decentralization, security and monitoring mechanisms. "Without a suitable governance framework and operating model, it is very difficult for businesses to fully exploit the value AI brings," Mr. Tuan said.

Additionally, the way AI's success is measured is also changing. According to him, in the past, businesses were often measured by the number of chatbots or AI applications deployed, but now AI is evaluated through business indicators such as labor productivity, project implementation time, customer experience or the ability to generate new revenue. Citing evidence from business reality, FPT representative said that many recent international contracts were signed not only because businesses have AI technology, but because they can commit to improving productivity and operational efficiency thanks to AI. For example, a large healthcare corporation in the US recently signed a four-year cooperation contract with the group, with a value about 30% higher than the previous contract, thanks to FPT's commitment to improve productivity by 20% through AI applications.

Race for human resources and organizational ability

If data and governance are the foundation of AI, humans are seen as the determining factor in the scalability of technology. A Microsoft representative said the company aims to train AI skills for more than 2.5 million people in the region, considering this a condition for AI to create real value instead of just stopping at technology deployment.

"I believe universalizing AI capabilities will make a difference," Mr. Mayank said.

According to Mr. Pham Minh Tuan, many businesses are still underestimating the role of humans in AI transformation. He compared AI to a "heavy weapon", but if the user is not competent enough, powerful tools will be difficult to be effective, and can even bring risks.

From implementation experience, FPT believes that businesses that want to successfully transform AI need to focus on three pillars. The first is to redesign business processes so that AI is integrated from the beginning, instead of adding to old processes. The second is human transformation. According to Mr. Tuan, the important skill in the AI ​​age is no longer just knowing how to use tools, but the ability to continuously collaborate between humans and AI in the same work process. The final pillar is the governance platform. As AI agents and digital human resources become more and more involved in business activities, the management model must also change to coordinate, monitor and connect humans and AI towards the same goals.

To realize these pillars, FPT also introduced CASAN - an AI-Native capacity assessment framework with five maturity levels: Curious - Augmented - Standard - Automatic - Native (Explore - Enhance - Standardize - Automate - Pure AI). This tool helps each business clearly determine where they are on their AI journey and what they need to invest in next, instead of expanding AI in a fragmented, directionless way.

"It's not that people do this step, AI does that step. Both need to interact in the same step to create optimal results," he said. This is also one of the reasons for the cooperation between FPT and Microsoft to increase the capacity to implement AI services in practice.

"The story does not stop at a certain technology being deployed, but rather about realizing a specific business result for customers," Mr. Mayank said. "Partners like FPT will play an important role. We need such a partner to truly transform what are inherently technological tools like AI into real results for customers."

Luu Quy