AI strategy for SMEs: three use cases, not a strategy paper
AI strategy for SMEs: what to tell the board
The board asks “what are we doing with AI?”, and the reflex is to commission a strategy paper. For an SME with 5 to 80 people, a useful AI strategy fits on one page: three concrete use cases, one owner in management, a few rules, 90 days, then a review. That gives the board something it can check in three months. A thirty-page paper usually gives it a vocabulary.
This is for owners and managing directors who have been asked the question, by the board, by a partner or by their own team, and who have no AI department to delegate it to. It is not about whether AI matters. It is about what answer changes the work.
Who this format fits
- Recurring office work and approved company tools
- External pressure without an in-house AI team
- One quarter to learn, not ten pilots at once
Who should start elsewhere
- Large firms with their own IT and data teams (this is only a starting point)
- Product or MVP questions (a different blog cluster)
- Team culture only, with no named process (workshop first)
Why strategy papers rarely change operations
A strategy paper describes a direction without naming a task, a person or a date, so Monday stays as it was. It lists opportunities per function, perhaps a maturity model and a vision. Then it is presented, approved and filed. Nothing in it says who does what differently next week.
Meanwhile the team keeps doing what it already does: drafting and translating in whatever chat is open, while the recurring work still runs on copy and paste. We have described that gap in everyone uses AI, operations still run on copy-paste. A paper does not close it, because the paper never touches the work.
BDO described the pattern in June 2026: large companies have already built AI into their value creation, while many SMEs stop at individual applications. Their advice: instead of launching as many pilots as possible at once, choose a few concrete use cases with measurable benefit and implement them consistently. And treat AI as a strategic task for management, not as an IT topic. That is BDO’s advice, and it is a good definition of a strategy that fits an SME.
Three use cases, chosen on purpose
Pick three tasks that already happen every week, each with a person who does them today and an effect you can see within 90 days. Three is enough to learn something and few enough to finish.
A mix works better than three of the same kind:
- One task everyone touches. Correspondence or translation in the approved company tool. The point is not speed alone. It moves drafting out of private accounts and under your rules.
- One recurring office process. The Monday report, invoice capture, quote follow-ups. It should repeat, hurt, have a clear start and end, and use data you already have. The filter is in which process to automate first.
- One task close to customers or revenue. For example first replies to standard enquiries, drafted by the tool and sent by a person. Visible to the board, and small enough to control.
For each case, write four lines: what changes, who owns the task, how you will see the effect, and by when.
- Three use cases
Weekly tasks, each with a named person.
- One owner
Someone in management reports on all three.
- A few rules
Approved tools, data classes, human review.
- 90 days
Work on the three. Nothing new gets added.
- Review
Keep, change or drop each case, then choose the next.
Shadow AI: why rules come before case one
Many firms already use AI, just not under company rules. Drafts in private accounts, customer data in free tools, no one sure which model sees what. Three official use cases do not fix that by themselves. They give management a frame: approved tools, data classes, human review before anything leaves the firm.
Without rules, the first “successful” case often spreads shadow use. With rules, you can tell the board that adoption and control belong together. The page for that is the AI usage policy template, and the tools part is how to approve AI tools.
One owner in management
The owner sits in management, because the decisions are about priorities and people, not about software. That person can free time for the three cases, say stop, and report to the board. An IT partner can handle settings. They should not decide which work changes.
The board rarely wants slides full of maturity stages. It wants to know who owns the three cases until quarter end and whether the rules are written down. One page with names, three cases and three dates often beats a consultant deck with twenty diagrams.
90 days, then a review
After 90 days each case gets one of three verdicts: keep it, change it, or drop it. Measure the work, not the logins. How long does the task take now, how often does it need rework, does it still run when the usual person is on holiday?
Counting logins is a common trap. The adesso GenAI Impact Report 2026 for Switzerland found that 45 percent of the Swiss companies surveyed measure GenAI success by usage rate, almost as many as measure process efficiency (47 percent). Their figures. A rising usage rate tells the board people open the tool. It does not tell it whether any work got better.
The board then gets one page: three cases, three verdicts, the next three. Repeat each quarter. That page is your AI strategy, and it stays current because it is rewritten every 90 days.
When three use cases are not enough
Some situations need more than one page. If you sell into the EU, have an establishment there, or use software that decides about people, such as screening applications, you need legal advice alongside the use cases. If a bank or an investor asks for a formal strategy document, write it, and put the three cases on page one.
And if your firm is far beyond 80 people with its own IT and data teams, this format is a starting point for one department, not for the company.
Does an SME need an AI strategy?
It needs an answer that changes work: three concrete use cases, one owner in management, a few rules and a review after 90 days. BDO advises SMEs to pick a few use cases with measurable benefit and make AI a management task. A long paper without tasks, people and dates rarely changes operations.
If you want to find the three cases with the team
If management wants to pick the three cases together with the people who do the work, and get everyone onto the same rules and tools, that is the AI in Business workshop, which is written for leaders asking exactly this question.
If one of the three is already obvious and is one recurring task, the Process Check looks at that task on its own.
Written by
Aurum Avis Labs
Builds and ships at Aurum Avis Labs. Writes here about what we learn working with founders and SMEs in the DACH region.
Related Articles
You might also be interested in these articles
Workshop or implementation: which one you actually need
AI workshop or Process Check: when the team needs shared rules, and when a single workflow should be built.
AI usage policy template for SMEs: a page you can copy
AI usage policy template for Swiss SMEs: eight clauses from approved tools to data classes and human review, ready to copy. Not legal advice.
Which process should an SME automate first?
Which process should an SME automate first? Four questions on frequency, pain, a clear end and data. The check is free.