The SME AI Trap: Why Waiting to “Get Ready” Is Costing You

AI implementation expert Patrick Hanhart has led more than 150 AI projects in SMEs, education, the social and medical sector, and industry. His core message: AI adoption is less a technology question than a people question. Here is why your team may feel stuck, where the real quick wins are, and how to bring everyone along.

1. Start with a conversation, not a tool

Asked what every SME should do with AI, Patrick did not name a tool. His answer: start talking about it. What worries your people? What do they expect? What do they wish for? And not only at leadership level, but at every level, from apprentices to the executive team. An open culture and plenty of communication help more than any software.

There is often a quiet hurdle on the leadership side. Many SME leaders are experts and craftspeople who know every screw in their production. AI is something they will never fully understand, and as long as they believe they have to master it alone, they won’t talk about it with their team. Patrick calls this a mindset hurdle. Admitting that you don’t know everything, and that your apprentice may know more than you do, is where progress starts.

2. The “we’ll do AI later” illusion

Patrick hears one sentence again and again: “We still have so many other things to do. Once that’s done, we’ll look at AI.” His counter-question: why not use AI to solve today’s challenges more easily?

Yes, AI gets better every day and keeps getting easier to use. But the learning curve has to start at some point, and it doesn’t happen overnight. Begin by engaging with AI, talking about it and running first projects. Your competitors have access to the same possibilities. Readiness comes from doing, not from waiting.

3. Stop looking for the “Zurich Airport” case

Patrick sees misplaced expectations as the most expensive beginner mistake. Leaders read about impressive examples, like the Zurich Airport assistant on WhatsApp that tracks your flight, reports delays and handles tens of thousands of requests per hour, and then ask where their own efficiency gains are.

For an SME, that logic rarely works. If I invest time and money in training and tools, where do I get it back? An airport processes thousands of requests an hour, an SME can’t. The better question: where can I create added value, and what can I do with the same resources thanks to AI?

He names a second costly pattern: treating AI as an IT project. Someone comes up with an idea from above, but nobody in the team wanted it, or there wasn’t enough critical mass for any leverage. That is why you need to take the whole department along from the start.

4. Where the quick wins actually live

Where AI pays off depends on your work. What is your core process? Which service would your customers like more of? What are your competitors doing with the same tools? From there, Patrick sees quick wins on several levels:

• Quality management: Maintaining the quality handbook is slow, tedious work. Every department head has to describe their processes, often at night or on weekends. With a few skills or bots set up centrally, everyone can contribute and reach a solid result fast, one that is attractive to read and actually used, instead of pulled out of the drawer for the next recertification.
• Project management: Preparing and leading projects, as more and more project work lands on organizations. Patrick’s practice tip: organize the Christmas party as a real project, with Claude or Copilot at your side. You learn project management and the tools at the same time.
• Strategy: Developing a vision or a strategy together with AI support.
• Support processes such as HR: Can a small HR team keep up in a field that moves this fast? Sometimes outsourcing is the smarter move. Ask which race you can actually win, and think carefully before jumping in just because everyone else does.

On tools: for an SME without an IT department, Patrick would start with Claude by Anthropic. Other tools are fine for free. Projects keep your prompts and chats per topic together, so you don’t start from scratch every time, and skills let you define, for example, how your PowerPoint should always look.

5. Make a plan, and stay agile

Whatever your size, you need a plan of your own. Where do you want to be in two years? What is the market doing, and what is still missing? Patrick’s starting point: run a survey in your team (what do people already do with AI, privately and at work, and where do they see opportunities?), develop a strategy together, and add a short market analysis for your industry.

Then separate simple applications from those that touch data and interfaces, and prioritize. Some things take two years, so start them now, and use simple applications to get the organization moving in the meantime. Decide consciously whether to invest yourself or wait for providers like Bexio or Abacus, who will add AI to their offerings sooner or later. Either way, watch the market, exchange with peers, and expect to adjust your direction about every six months.

6. The human side: rules, trust and learning together

AI touches people, so adoption is never only about software. Leaders are responsible for what their teams do, and employees expect clear rules in return.

1. Set clear data boundaries: For non-sensitive data, meaning no patient data, no employee data and no trade secrets, people can use the tool that suits them best. For sensitive data, build a protected environment or use secure interfaces, for example to OneDrive, so nobody has to be afraid of working with it.
2. Build trust and train together: You can ban everything, and people will still use AI on their private laptops. So train together, look each other in the eye and make sure everyone understands why something is not allowed. It never ends with one training, because tools change every three to six months. Individual learning paths work better than one training for all: management needs to understand the value for company and customers, employees how a tool works, so they can give feedback.
3. Take the slowest along: A team moves at the pace of its slowest member. Generations bring different experiences and fears, and you can’t tell someone at the annual review that one colleague worked with AI and the other without, and both delivered the same performance. Address fears actively: explain why you look at AI, where the opportunities and risks are, and what it means for each person. Skeptics have fair points, from geopolitics to children growing up with AI, but the topic can’t be left out. The more people know, the less they fear. Small successes help: let them see why an answer was bad, why another was good, and what their own influence as a human is.
4. Learn from each other: A colleague who shows that a task took half an hour instead of one convinces skeptics far better than an external expert saying it’s easy. Put AI on the agenda of your monthly team meeting, plan time for it, and let the managing director ask the apprentice about that clever prompt. An AI inventory helps too: not only for control, but to see where AI is used and where people need support.

7. One piece of advice for leaders: Stay curious

Patrick’s one piece of advice for an SME leader: be curious. Especially about the hidden talents in your team. Someone may suddenly show up whom you never expected to be interested, who is AI-savvy and can help your whole team along the way.

Find these people, give them room to experiment and let them become your AI ambassadors. Patrick also sees a shift in hiring: he would rather take someone who is already at a certain AI level than someone with strong skepticism who first has to be convinced.

This article is based on a LinkedIn Live session on 30 September 2026, hosted by v-oice founder Aileen Zumstein with v-oice AI implementation expert Patrick Hanhart.