AI is becoming a ‘capability asset’ as businesses train it and integrate it into their workflows. Photo: AI
AI is no longer a free tool for experimentation, but it has become a form of ‘capability capital.’
To acquire it, both businesses and workers are paying with money, time, and a fundamental reorganization of how work gets done.
On a Monday morning in Ho Chi Minh City, employees at AstraZeneca Vietnam begin their day by logging into AI assistants they have built themselves.
When data are retrieved, answers are synthesized almost instantly.
Meetings that once stretched to an hour now wrap up in under 20 minutes.
Nguyen Hung Cuong, a category manager in cardiovascular, renal, and metabolic care at AstraZeneca Vietnam, shared that he does not consider himself deeply technical.
Yet through internal training programs, he successfully built an AI assistant tailored to his work.
He trained the system using the company’s own assets–standard operating procedures, approved internal documents, and specialized product knowledge–effectively creating a compact ‘knowledge hub.’
When queried, the AI assistant retrieves relevant materials and delivers accurate, consistent answers in seconds.
Previously, searching for documents or verifying information could take hours, but now it happens almost instantly, he said.
However, the real value lies beyond speed.
The assistant is shared across teams, especially with new hires.
Instead of spending weeks onboarding, employees can be supported by the AI and access institutional knowledge immediately.
Quietly, workplace culture is shifting, from reliance on individual experience to dependence on shared knowledge systems, with AI acting as the intermediary.
Rather than relying on generalized internet knowledge, companies are turning AI into a kind of ‘internal employee’ by connecting it directly to proprietary data.
Systems retrieve relevant documents first, then generate responses grounded in company-specific context.
Atul Tandon, CEO of AstraZeneca Vietnam, said that the company has built AI platforms at a global scale, along with structured internal training systems.
Beyond accelerating research and development, this approach promises faster insights, smarter experimentation, stronger collaboration across healthcare and technology ecosystems, and accelerated digital transformation.
Local firms enter the fray
This shift is not limited to multinational corporations.
Domestic firms are joining the race in their own ways.
Mai Duc Hoa, founder of Dr Media based in Saigon Ward, Ho Chi Minh City, shared that his company has invested heavily since early 2026 to send employees to AI training programs led by international experts.
In marketing and digital content production, AI does more than assist–it can fundamentally reshape the role of a chief marketing officer when used effectively, he said.
From ideation and scripting to video production, AI can execute entire workflows with just a few prompts.
His firm is experimenting with a range of tools.
Video platforms such as Kling AI, Veo, and Higgsfield AI are used to generate clips from scripts, while HeyGen produces AI-generated voices and avatars for media content.
Early results showed significant impact, with some videos and content reaching millions of views.
Yet Hoa stressed that the effectiveness lies not in the tools themselves, but in the user.
AI can produce videos quickly, but without a deep understanding of storytelling and craft, the output lacks identity, tone, and emotional resonance, he said.
As such, the company emphasizes not just tool proficiency, but also a deeper understanding of content creation and how to direct AI effectively.
AstraZeneca Vietnam employees at a sharing session on the process of ‘training an AI assistant.’ Photo: Supplied
Workers invest to stay competitive
If companies are investing at the organizational level, workers are competing on a more personal and immediate front.
Ngoc Minh, a data analyst in Ho Chi Minh City, has spent around VND50 million (US$1,900) on an advanced AI course, a significant sum relative to his income.
In return, he learned not just to use AI, but to design systems around his work.
He built a workflow that automatically collects, analyzes, and synthesizes data from multiple sources.
His ‘assistant’ is not an available product but a combination of generative AI models such as ChatGPT, integrated with company data and his own eight years of accumulated materials.
Each morning, the system prepares a summary report in advance, eliminating hours of manual data review.
To further enhance it, Minh combines multiple tools, including automation platforms like Zapier and development frameworks such as LangChain.
Together, they form a closed-loop process: data is collected, processed, analyzed, and then synthesized into recommendations.
The result is not just improved productivity, but a shift in role, from executor to system designer.
Booming training market
As demand surges, AI education has rapidly become a premium market.
In the United States, programs like ‘Leading in Artificial Intelligence’ at Harvard Kennedy School cost around $11,000 for a one-week, in-person course combining policy theory with real-world case studies.
Meanwhile, the ‘AI Strategy and Leadership’ training program from MIT runs for 12 weeks online at some $7,750, focusing on integrating AI into organizational strategy, from data infrastructure to risk management.
Shorter courses on platforms like edX typically cost between $2,500 and $3,500.
In Vietnam, the trend is emerging with a more practical orientation.
Programs such as ‘Innovative Leadership with AI’ by the PACE Institute of Management focus on operational application, guiding leaders to directly use tools like ChatGPT, Claude, and Zapier to automate workflows, analyze data, and support decision-making.
Building AI-driven operating models
Huong Lan, an operations manager at a corporation in Ho Chi Minh City, said that the real difficulty lies in ensuring that hundreds of employees can use AI tools efficiently, safely, and consistently.
In the strategy courses she has attended, with some costing up to $5,000, the focus is not on coding, but on designing an ‘AI operating architecture.’
Tools like Microsoft Copilot are not used freely, but are tightly integrated with internal data sources stored on SharePoint.
Once processed by AI, data do not simply answer questions.
They feed into analytics platforms such as Power BI, generating real-time dashboards.
In this model, AI synthesizes and suggests, while users retain final decision-making authority.
The key is not individual tools, but how they are connected into a seamless system that enables faster and more accurate decisions, Lan emphasized.
