What Small Businesses in Vietnam Actually Use AI For
This publication is publicly available with an advisory disclaimer but has not yet received recorded CLO approval. Verify current law and procedure before relying on it.
General information only, current at the time of writing. Rules for foreigners and small businesses in Vietnam change - please confirm your own situation before acting on it.
Direct answer
Most small businesses in Vietnam have tried AI; considerably fewer have changed how they work because of it. Adoption is wide and shallow. The tasks that persist are language tasks - drafting, translating, replying - and the businesses that achieve more than that are the ones that documented their own processes first.
The constraint is rarely the technology or even the cost. It is that a tool cannot be directed at a bottleneck nobody has described.
The shape of adoption
Industry surveys published in 2025 suggested that a large majority of Vietnamese companies had adopted AI in some form, while only a small minority had deployed it at any meaningful scale.
That pattern - near-universal trial, rare embedding - describes the market more usefully than any single figure, and it matches what we observe across small clients.
The relevant question is therefore not whether to adopt but what distinguishes a business where these tools are a curiosity from one where they removed an actual recurring task. That distinction turns out to have little to do with which tools were selected.
What actually gets used
Language work, by a substantial margin. Drafting customer replies. Translating between Vietnamese and English, Korean, Japanese or Chinese. Writing product descriptions. Correcting a message before it is sent.
These persist for an identifiable reason: the benefit is immediate, the risk is low, and the output can be judged. A translation is either usable or it is not, and the person using it can tell within seconds.
A second tier is less universal but growing: summarising long message threads, drafting standard documents from a template, and providing first-line answers to repetitive customer questions. These share the same characteristics - language-based, verifiable, and low consequence if wrong.
What is tried and quietly abandoned
Anything asked to replace judgement rather than to reduce typing.
Pricing decisions. Resolving a complaint where the customer's history matters. Anything depending on knowledge of this particular supplier, this particular relationship, this particular exception. Tools produce confident answers in these areas, and the answers are wrong in ways that require experience to detect.
The abandonment is usually silent. Nobody announces a decision to stop; the tool is simply used less, then not at all. Because the decision is never articulated, the business does not record what it learned, and something similar is attempted six months later with the same result.
Recognising this as a boundary rather than a failure is more useful than it sounds. Language tasks and judgement tasks behave differently, and a business that knows which is which can direct its effort accordingly.
Why cost is cited but is not the constraint
Surveys consistently place cost at the top of the list of obstacles, with a shortage of internal skills close behind. Both are genuine constraints.
The amounts involved for a small business are nonetheless modest relative to rent, wages or stock. Where cost is the stated reason and the sums are small, it generally signals something else: uncertainty about whether the expenditure would produce anything.
That uncertainty is reasonable. A business that cannot identify which task would become faster genuinely cannot judge whether a subscription is worthwhile, and declining to spend on an unmeasurable benefit is sound commercial judgement rather than reluctance. Cost becomes the language for an unresolved question about value.
What separates the businesses that benefit
One factor more than any other: they documented how their own processes work before selecting tools.
| Business that benefits | Business that does not |
|---|---|
| Knows which task consumes the most time | Experiments broadly, measures nothing |
| Can describe how information moves | Holds it all in individual heads |
| Selects one bottleneck and addresses it | Trials many tools shallowly |
| Keeps personal data out by default | Enters whatever is to hand |
| Reviews whether it helped | Moves on to the next tool |
Nothing in the left column is technical. It is ordinary operational clarity, and it is what converts a tool into a result. A business with that clarity can evaluate any new tool in an afternoon; a business without it cannot evaluate any tool at all, regardless of how long it spends.
Where this is heading
Two movements, both worth planning around.
Tools are becoming cheaper and easier to operate, which means the cost barrier will continue to erode and the skills barrier will narrow. What will not narrow is the need to understand one's own processes, and that will become the main factor separating businesses.
At the same time, expectations around data handling and automated systems are tightening. Casual use of customer information - which many small businesses have practised without particular thought - is becoming a decision that has to be made deliberately rather than by default.
The sensible response to both is the same and unglamorous: write down how information actually moves through the business. It identifies what to automate and it identifies where the obligations sit, and it costs an afternoon.
Frequently asked questions
How widely are small businesses in Vietnam using AI? Industry surveys published in 2025 suggested that a large majority of Vietnamese companies had adopted AI in some form, but only a small minority had deployed it at scale. That gap between trying and embedding is the most striking feature of the picture.
What do small businesses use it for most? Language work, overwhelmingly - drafting messages, translating between Vietnamese and other languages, writing product descriptions and replying to customers. These are low-risk, high-frequency tasks where the benefit appears immediately.
What holds small businesses back? Surveys point to cost as the most commonly cited barrier, followed by a shortage of internal skills. In practice a third factor matters just as much: very few businesses have written down how their own processes work, which makes anything beyond casual use difficult.
Where does AI disappoint small businesses? Anywhere it is asked to replace judgement rather than reduce typing. Pricing decisions, customer disputes and anything requiring context about a specific relationship tend to produce confident answers that are subtly wrong.
What separates businesses that benefit from ones that do not? Whether they wrote down their own process first. Businesses that know how information moves through their operation can point AI at a specific bottleneck. Businesses that have not end up experimenting endlessly without anything changing.
Conclusion
Adoption in Vietnam is wide and shallow, which means the opportunity sits in the gap between trying and embedding rather than in access to tools. Language tasks persist and judgement tasks are quietly abandoned, and recognising that boundary saves repeated effort. And the businesses obtaining real value described their own processes before shopping.
For assistance identifying the single bottleneck worth addressing first, talk to LotusGuard. We do not resell software, so the recommendation is not the product.
