powered_by-logo reporter-logo inbusiness-news-logo GOLD-DIGITAL-EDITIONS

Small businesses need an AI strategy before their employees write one for them

Almost every business owner I speak to in Cyprus asks me the same questions about AI. Where do we start, and what should the roadmap look like? Where will the return come from? How do we keep our data safe once AI starts acting on our behalf? Nobody I meet is against AI but there is scepticism. Most owners are simply waiting for clear answers, and while they wait, the rest of Europe moves ahead. According to Eurostat, fewer than one in ten Cypriot enterprises used AI in 2025, less than half the EU average. A recent Eurobarometer survey of small and mid-sized companies points to the same conclusion: only 7% of those in Cyprus use AI extensively, against 12% across the EU, and 67% cite a lack of in-house expertise as the most common obstacle.

Your people have already started

In most companies, people are already using AI, often on personal accounts the business cannot see. Microsoft’s Work Trend Index found that 78% of people who use AI at work bring their own tools, rising to 80% in small and medium-sized companies. The industry calls this shadow AI. With good intentions, people paste in customer details, pricing and internal documents to get their work done faster. And it isn’t just junior staff. I regularly meet executives who run much of their working day through a personal AI account, putting some of the company’s most sensitive information outside its control. The longer a business waits to define its approach, the more of its data ends up in tools it can neither see nor control.

This is why I believe even a company of fifteen people needs a proper AI strategy, with security built in from the outset. It can be short. What matters is that it defines which tools are approved, what data they can access, who owns the results and how those results will be measured.

The stakes rise as AI gains more agency, moving from answering questions to acting on its own. An agent that updates your CRM or writes to a customer needs clear limits on what it can access and send. Without a plan for those limits, agents will multiply the risks that shadow AI has already created.

 

Sounding like everyone else

A second risk is one I suspect every reader has seen in their own inbox. Companies that hand their voice to a generic tool end up sending the same polished, interchangeable messages as everyone else. Readers recognise this kind of AI slop and have little patience for it, while a small business that loses its identity gives away one of the few advantages it holds over larger competitors. Everyone has access to the same models, so what keeps your output distinctive is what only you know about your customers. An agent is only as good as the context you give it.

 

Start where the return is most visible

So where should a small business begin? My answer is close to revenue, which for most companies means sales. That return is not automatic. In PwC’s 2026 Cyprus CEO Survey, only 22% of chief executives said AI had increased revenue over the past year, which is why I would start where the gain is easiest to see. Sellers still spend most of their week on work other than selling: Salesforce’s latest State of Sales report puts actual selling time at around 40%. Owners also tell me repeatedly that they cannot find good salespeople or keep them once they do.

AI can be a force multiplier for the team you already have. Agents can research accounts, prepare meeting briefs, draft follow-ups and keep the CRM up to date, giving sellers more time to spend with customers. The bigger prize is what a small team could never do before, such as reaching every customer in a far more personal way each week. And because sales is already measured, the return shows up in numbers every owner follows, such as the sales pipeline and win rates.

 

Where good projects go wrong

Choosing the right place to start does not guarantee a good result. Many technology projects begin with real promise and still disappoint, and in my experience, the cause is rarely the technology itself. Business owners know their companies inside out but often find it hard to translate their needs into clear technology requirements. The projects that work are the ones where someone asks the hard questions early, before any tool is chosen or any partner is briefed.

 

Building the right foundations

AI amplifies whatever system it sits on. This is why businesses should begin by examining their sales operating model, focusing on processes and people: who they sell to, how a deal moves, who owns what and how their teams are coached and rewarded.

The next step is to assess how ready the business is for agentic sales, identifying strengths, gaps and areas for improvement. From there, businesses can develop a clear AI and sales strategy based on agreed KPIs, including which tools are approved and what data they may access. Teams can then gradually hand more work to AI agents, with human oversight maintained for anything that reaches a customer.

Once the requirements are properly defined, businesses can bring in specialised technology partners in areas such as infrastructure, data, security and AI. Reliable data and modern infrastructure allow companies to adapt quickly, and partners do their best work when the important questions have already been answered.

 

Taking the first steps

Fear should not drive adoption, and neither should the pressure to buy the latest tool. A good first step is to find out which AI tools your people already use, leadership included. Then approve one that works safely on your own data and explain why. Banning AI doesn’t stop people from using it. It just pushes it out of sight. From there, pick one or two use cases close to revenue and agree, before you start, what results you expect and who will be responsible for achieving them.

Ultimately, the goal should be to help businesses of all sizes become more efficient and achieve things they could not do before, while ensuring that their customers feel the difference.

 

*Nicolas Yiallouros is Co-Founder of Mirai, specialising in AI-driven sales transformation. He has extensive experience in enterprise technology sales at Bloomberg, Salesforce and Microsoft, with expertise in security and data governance.