Have you ever struggled to hire AI engineers in Laos? We hit the same wall. However, after designing a training program that gradually builds AI literacy among existing staff, we were able to produce talent from within the company capable of applying AI to real business operations within 6 months. In this article, we walk HR and training professionals at companies operating in Laos through a step-by-step guide on how to design and implement a 3-stage AI literacy training program targeting non-engineers, complete with actual measured data from our own experience.

Understanding the structure of Laos's IT talent market reveals why the approach of "hiring AI engineers" tends not to work well.

When I launched my first project in Laos, I posted a job listing for Python engineers and received zero applicants. The annual number of graduates from the IT faculty at the National University of Laos is limited, and many of them end up at companies in Thailand or Vietnam. Given the wage disparities within ASEAN, this is an entirely natural outcome.

Taking stock of the ICT talent landscape in Laos, the following structural challenges come into focus:

• Shortage in absolute numbers: The number of IT-related graduates falls far short of domestic demand
• Intra-regional brain drain: Companies in Thailand and Vietnam poach talent by offering salaries two to three times higher
• Near-zero AI specialists: There are almost no professionals in the country specializing in machine learning or data science
• The English proficiency barrier: The pool of talent capable of reading technical documentation in English is limited, constraining the pace of learning

Comparing recruitment costs versus training costs, the ROI of training is overwhelmingly higher in Laos.

Based on our experience, the annual cost of bringing in an engineer from overseas was more than five times the training costs for three local staff members. Moreover, the recruited engineer left after one year. In contrast, locally trained staff have a deep understanding of the business context, which means their proposals for leveraging AI tools are well-targeted, and their retention rate is high.

Before diving into training design, three prerequisites need to be in place. Skip this step, and the training will end up being nothing more than a "let's casually play around with ChatGPT" session.

AI literacy training takes up 2–4 hours of participants' working hours per week. Without explicit approval from senior management for "training participation during working hours," frontline managers tend to be reluctant to allow attendance. At our company, the CEO declared at the kickoff meeting that "AI utilization is a company-wide strategy," which caused the participation rate to jump from 60% to 95%.

Instead of "learning AI," make the goal "solving this business challenge with AI." Before training begins, take stock of each department's business challenges and identify 3–5 tasks that can be improved with AI.

Inventory perspectives:

• Repetitive tasks: Work that is repeated daily or weekly (data entry, report creation, email correspondence)
• Standardizing judgment: Tasks that require individual judgment but follow somewhat predictable patterns
• Language processing: Tasks involving translation or summarization between Lao ↔ English ↔ Japanese

The first stage of the three-stage training design is to help participants understand the basic concepts of AI and "what it can and cannot do."

Target: All staff (engineers and non-engineers alike) Frequency: Once a week × 2 hours × 8 sessions (16 hours total) Format: Workshop-based (lecture-style instruction kept to 30% or less of total time)

Using English technical terms as-is significantly reduces participants' comprehension. Our company implemented the following measures.

Localizing terminology into Lao: We created a glossary that paired basic terms such as "prompt," "hallucination," and "token" with Lao-language explanations, which was distributed on the first day of training. This drew on the know-how we had accumulated when developing our Lao-language AI chatbot.

Smartphone-first: Smartphone penetration in Laos is higher than that of PCs. The hands-on training sessions were designed to be completable using the ChatGPT app on a smartphone, so that no staff member would be excluded from participation due to not having a PC.

Pair learning: We paired participants with differing levels of IT literacy, with the more proficient participant taking on a teaching role. In addition to the effect of deepening understanding through teaching, this approach aligned well with the Lao culture of mutual support, and significantly reduced the dropout rate.

In the second stage, the fundamentals learned in the first stage are applied to actual work tasks. This is the core of the training.

Unlike the foundational training common to all employees, this phase divides the curriculum by department.

The most effective approach in our company's training was a structure that minimized classroom learning and centered on hands-on practice in real work.

• 70%: AI utilization in actual work (integrating AI into day-to-day tasks)
• 20%: Feedback from mentors and colleagues (weekly retrospective meetings)
• 10%: Classroom learning (introduction of new techniques and case studies)

In the weekly retrospective meetings, participants share "what they were able to solve with AI this week" and "what they delegated to AI but didn't work out." Sharing failure cases is particularly important — when one accounting staff member reported that "the figures in a financial summary generated by AI were incorrect," it instilled a team-wide habit of always verifying AI output.

To avoid ambiguity in training outcomes, clear evaluation criteria are established.

The goal of Phase 3 is to cultivate at least one "AI Promotion Leader" in each department. There is no need for everyone to become an expert. If there are individuals who can drive AI adoption within their department and serve as a resource for their colleagues, the overall level of AI utilization across the organization will improve autonomously.

From the Phase 2 evaluation results, select staff who meet the following criteria:

• AI usage frequency in the workplace ranks in the top 20% of the team
• Actively shares failure cases and is able to propose improvements
• Has no reluctance to teach colleagues

At our company, 4 AI Promotion Leaders were selected from 15 participants. Interestingly, 3 of the 4 came not from the IT department, but from Sales and Customer Support. A mindset of "let's solve business challenges with AI" proved more important as a leadership quality than technical aptitude.

The following additional training will be conducted for AI Promotion Leaders, separate from general participants.

API Integration Fundamentals (4 hours × 2 sessions): Learn how to integrate AI APIs into business workflows using no-code/low-code tools (Zapier, Make). No coding is required, but participants will develop an understanding of the concept of "incorporating AI into automation pipelines."

Applied Prompt Engineering (2 hours × 3 sessions): Practice advanced prompt techniques in real-world settings, including few-shot prompting, Chain of Thought, and output format specification.

Training Facilitation (2 hours × 2 sessions): Practice teaching methods and workshop facilitation so that leaders can lead training sessions for the next cohort of participants. This enables the internalization and sustainability of training programs.

Numbers speak louder than words. Here we share the results of the AI literacy training conducted at our company's local subsidiary in Laos.

• Participants: 15 (Sales: 4, Accounting: 3, Customer Support: 4, HR: 2, Logistics: 2)
• Duration: 6 months (Phase 1: 2 months + Phase 2: 2 months + Phase 3: 2 months)
• Training Cost: AI tool licensing fees + man-hours for training design and management (no external instructors used)

The most dramatic change occurred in the Customer Support department. Before the training, English inquiries were handled manually one by one, but after the training, the team transitioned to a workflow in which AI generates draft responses that staff then review and revise. Not only did the response speed more than triple, but the quality of English also improved, leading to better customer satisfaction scores.

While it appeared to be progressing smoothly, there were two major failures along the way.

Failure 1: Starting with English-only materials. During the first two weeks, training began using only English materials, and average scores on comprehension tests fell below 40%. Supplementary materials in Lao were hastily created and a glossary was distributed, after which average scores recovered to 75% on the next test.

Failure 2: Continuing the shared curriculum for all departments into the third month. At the point when we should have moved into Phase 2, we continued with a shared "advanced module" for everyone, which triggered complaints that the content was "irrelevant to my actual work." We quickly switched to department-specific curricula and incorporated job-related tasks as practice material, after which motivation made a V-shaped recovery.

In addition to our own failures, we will share anti-patterns observed from cases of other companies in the ASEAN region.

A pattern where teaching "how to use ChatGPT" becomes the goal itself, failing to lead to the resolution of actual business challenges. Participants say they "found it interesting," but a month later, no one is using AI.

Workaround: Set the training KPI to "reduction in time required for business tasks" rather than "mastery of tool operation." Measure specific Before/After results for actual business tasks on a weekly basis.

The IT department is technically proficient, but does not have a deep understanding of the business processes in sales or accounting. IT-led training tends to skew toward the technical side, causing non-engineer participants to fall behind.

Workaround: Training design should be led by HR and department managers, with the IT department taking a supporting role for technical matters. Build a structure where "business professionals learn how to use AI."

Incidents involving the input of confidential information into AI tools tend to occur immediately after training begins. "Deciding later" will be too late.

Workaround: Clearly communicate the following rules on the first day of training.

• Do not input customers' personal information into AI
• Anonymize financial data before use
• Business use on non-business accounts (personal accounts) is prohibited
• Always have a human review AI output before sending it outside the company

In our case, the main costs for training 15 employees were the business account fees for AI tools (monthly fee × number of users × 6 months) and the internal man-hours allocated to training design and administration. We did not use any external instructors; instead, senior in-house staff doubled as trainers. The total cost came to less than one-fifth of what it would have cost to hire a single AI engineer from overseas.

Can. In fact, since that is the majority in Laos, supplementary materials in Lao are essential. The AI tools themselves support input and output in Lao, so practical results can be obtained even when prompts are written in Lao. However, since resources for advanced prompt engineering are primarily in English, it is desirable for AI-driving leaders to have at least a basic ability to read English.

However, this does not mean that "employees won't resign if they aren't trained." In fact, the risk of employees quitting out of frustration due to stagnating skills is far greater. At our company, the one-year retention rate for training program graduates stands at 86%, surpassing the company-wide average of 72%. Creating a work environment where employees can apply their AI skills is the most effective retention strategy.

The three-tier structure can be used as-is, but "department-specific curriculum" should be reread as "individual curriculum." For teams of five or fewer, it is more effective to individually interview each member about their work challenges and set a practical theme tailored to each person. In fact, smaller teams have greater flexibility in adjusting the curriculum and tend to produce results more readily.

The most realistic way to secure AI talent in Laos is to develop existing staff. Below is a recap of the three-phase training framework outlined in this article.

1. Phase 1 (Months 1–2): Building foundational AI literacy. Targeting all staff. Workshop-style sessions allow participants to experience firsthand what AI can and cannot do.
2. Phase 2 (Months 3–4): Applying AI to daily operations. Department-specific curricula integrate AI into practical tasks, with a 70:20:10 distribution that emphasizes hands-on practice.
3. Phase 3 (Months 5–6): Developing AI champion leaders. At least one person is selected from each department to take ownership of internalizing and sustaining the training program.

The first step is to take stock of the operational challenges within each department. If you can identify three tasks that AI could improve, you already have everything you need to get started.

For the overall AI adoption framework, refer to the "AI Adoption Guide for Lao Businesses." For the technical foundation behind the chatbots covered in the training, see "How to Build an AI Chatbot That Supports the Lao Language."