Over the past few months, we spoke with several small and medium-sized businesses that all said roughly the same thing: “We know we need to do something with AI, but we don’t really know where to start.” Sometimes this came from curiosity. Sometimes from pressure from competitors. And sometimes simply because AI seems to be everywhere. What stood out was that almost no one had defined a concrete problem they wanted to solve with AI. And that’s exactly where things often go wrong.
In this blog, we share how we look at AI within small and medium-sized businesses:
- Where AI truly adds immediate value within your company
- Where AI is often not the right solution
- Why custom AI and process-driven thinking make the difference
No hype—just a realistic and strategic perspective.
The common mistake: AI as the starting point
Where many companies go wrong is starting with AI based on a feeling:
“Our competitor is using AI, so we should too.”
Choosing this as a starting point often leads in practice to:
- Disconnected tools that work alongside each other but don’t integrate or communicate
- Experiments without a clear owner or responsibility
- No measurable impact on time, cost, or quality
In those cases, AI doesn’t create a more efficient workflow—it adds an extra layer of complexity.
The question that should almost always be asked first is:
Which process within our company consistently costs time, money, or leads to errors?
Only when that is truly clear does AI become interesting.
Where AI does add direct value
In practice, we see that AI works best and most efficiently in areas where processes:
- Involve many repetitions
- Are data-driven
- Regularly contain human errors
- Take a disproportionate amount of employee time
These are often not the most exciting processes—but they are the ones with the greatest positive impact.
Customer communication & support
You might recognize this:
- The inbox fills up daily with customer questions
- Emails are forwarded multiple times internally
- Employees spend a lot of time answering the same questions repeatedly
AI can support this very efficiently by:
- Classifying incoming questions
- Automatically retrieving relevant information from, for example, a knowledge base
- Preparing draft responses
The result is not “impersonal communication,” but instead:
- Much faster responses to your (potential) customers
- Consistency in your answers
- Less pressure on the team
- An even better customer experience
So not by replacing people—but by relieving them.
Quotes, requests, and intake processes
At many small and medium-sized businesses, requests come in through multiple channels, such as website forms, emails, phone calls, or WhatsApp.
What typically happens next:
- Someone manually reads everything
- Information is interpreted
- Data is re-entered
All of this is done manually—and that’s understandable. Every request is unique and requires new calculations. AI can make this process much smarter and more efficient by:
- Automatically analyzing requests
- Structuring essential information
- Consulting a knowledge base with past requests
- Preparing a proposal (or next steps)
This leads to shorter turnaround times, which in turn results in higher conversion rates—without the team having to work harder.
Internal reporting and insights
Many companies have data—lots of data. Sometimes they even have data whose value they don’t fully understand yet. As a result, its use remains limited:
- Reports take a lot of time
- Insights arrive too late
- Decisions are partly based on gut feeling
How AI can help here:
- Retrieving and summarizing data
- Identifying anomalies and alerting you to them
- Automatically delivering reports or insights
The real value doesn’t lie in beautiful dashboards, but in reliable data that enables faster and better decision-making.
Where AI is usually not the solution
Now that we’ve outlined where AI can work well, it’s equally important to mention where AI currently does not work well.
Poorly defined processes
AI implementations rely on context. If that context isn’t clear or lacks structure, the process simply isn’t ready for AI yet.
If a process:
- Works differently every time
- Has no fixed structure
- Depends on improvisation
Then AI is more likely to reduce efficiency rather than improve it. The order should always be: first structure, then automate.
Complex human decision-making
AI is extremely strong at recognizing patterns and processing data, but very weak at:
- Emotion
- Politics
- Non-measurable context
Decisions about culture, relationships, and positioning remain human work.
Projects without a concrete goal
Using AI “because you can” often leads to:
- Low adoption within teams
- Resistance to change
- Disappointment with the results
The best AI projects don’t start with technology, but with friction within the business—an issue that repeatedly occurs in a specific process or task.
Why custom solutions almost always make the difference
No two companies are the same. Yet many organizations try to force their processes into standard AI tools. Sometimes that works—but often it doesn’t.
Standard tools:
- Are based on averages
- Only partially fit specific processes
- Force you to adapt your way of working
- Are inconsistent due to varying prompts
Custom AI completely reverses this:
- The solution adapts to your processes
- Integrates with your existing systems
- Focuses on one concrete problem and solves it in depth
That’s exactly why custom solutions often deliver results faster in practice than standalone tools.
Finally
AI is not a goal in itself.
It is a strategic layer on top of your business processes. When applied in the right places, with a clear objective and realistic expectations, it can have enormous impact.
But it always starts with understanding:
- Where you currently lose time, money, or quality
- Which processes are suitable for automation
- And where AI is not the right choice
