Artificial intelligence is evolving at an incredible pace. Almost every week, new models and capabilities emerge, making tasks that seemed impossible just a few years ago increasingly accessible.
There's undoubtedly a lot of hype surrounding AI. But I believe something important gets overlooked: we still don't know how far this technology can go.
And that's exactly why businesses should start experimenting with it today.
Not because every company needs an AI agent, but because understanding what AI can and cannot do is becoming increasingly valuable.
The Real Opportunity Is Experimentation
I genuinely believe AI agents will fundamentally change how businesses operate.
We're living through a period where developers and companies are constantly pushing the boundaries of what's possible. Some experiments succeed, others fail, but both teach us something valuable.
As AI models improve, tasks that are difficult to automate today may become straightforward tomorrow.
Companies that experiment now will be better prepared to recognize and take advantage of those improvements.
This doesn't mean investing thousands of Swiss francs in complicated AI projects. Sometimes, the best way to start is by identifying a small, repetitive task that nobody particularly enjoys doing.
And building something to solve it.
A Real Example: Automating Invoice Processing in Four Hours
I recently developed a small AI-powered automation for a hostel.
The problem was simple: supplier invoices arrived through email and other channels, and someone had to manually enter their information into the accounting system.
It wasn't particularly difficult work, but it was repetitive and time-consuming.
So we automated it.
The workflow is straightforward:
Incoming invoices → Shared folder → AI document processing → Accounting entries
Invoices automatically arrive in a designated folder. An AI-powered system reads the documents, extracts the relevant information, and generates the corresponding accounting lines.
The automation is already running in production and being used in daily operations.
The initial development took approximately four hours.
Four hours to automate a repetitive task that previously required manual data entry.
It also reduced our dependence on external document-processing services that charge per invoice.
Of course, not every automation can be built this quickly. More complex projects require additional testing, security measures, and integration work.
But that's precisely what makes this example interesting.
AI doesn't need to solve a massive problem to create real business value.
Sometimes, a relatively simple solution can make a meaningful difference.
From Business Problems to Working Prototypes
One of the biggest changes I've experienced as a software developer is how quickly we can now turn ideas into working prototypes.
With AI-assisted development, a small, well-defined automation can sometimes be prototyped in just a few hours.
That changes how I think businesses should approach innovation.
Instead of spending weeks discussing whether an idea might work, why not build a small prototype and test it?
The first step is understanding the problem.
Most business owners and employees already know which tasks consume unnecessary time. It might be processing documents, copying information between systems, responding to repetitive requests, or preparing reports.
Once we identify the problem, we can explore the simplest solution.
Sometimes that involves an AI agent. Other times, traditional business process automation or a software integration is more appropriate.
For example, an AI agent connected to an ERP or CRM could interpret incoming customer requests, retrieve relevant information, and prepare actions for an employee to approve.
But if two systems simply need to exchange structured data, a conventional API integration might be more reliable.
The goal isn't to use AI everywhere. It's to use the right technology to solve the right problem.
And importantly, a working prototype isn't necessarily a production-ready system. Reliability, data protection, error handling, and appropriate human oversight still matter.
AI Automation in Switzerland: Why SMEs Should Start Small
For small and medium-sized businesses in Switzerland, AI automation doesn't have to begin with a major digital transformation project.
It can start with a single process.
An invoice that needs processing. A customer request that requires information from multiple systems. A repetitive administrative task that consumes an employee's time every week.
These are practical opportunities to experiment, measure results, and learn.
I also use AI in my own daily workflow. Even this article began with me recording my thoughts rather than writing everything manually. AI helps transcribe and structure those ideas, allowing me to communicate my own experiences and opinions much faster.
It's another small example of how technology can remove friction from everyday work.
We don't know exactly what AI agents will be capable of in two or five years.
But I'm convinced that experimenting today is one of the best ways to prepare for what's coming.
Start small. Test your ideas. Understand the limitations. Keep improving.
The technology will continue evolving, and so will the opportunities to use it.
Have a Business Process Worth Automating?
I'm Sebastián Ignacio, a software developer based near Zürich, Switzerland, specializing in AI automation, ERP/CRM integrations, and custom software development.
I help businesses explore practical ways to reduce manual work and connect their existing systems.
Whether you're based in Zürich, Zug, or elsewhere in Switzerland, you don't need a massive AI project to get started.
Sometimes, all it takes is identifying one frustrating task and building a better way to handle it.