According to the Level Nord Blog (June 2026), 41 percent of German companies are now actively using AI – double the figure from 2025. The pressure to introduce a chatbot right now is palpable. And that is exactly the problem: many businesses buy in a rush, without asking themselves the right questions. The bill comes later – in the form of systems nobody uses, data protection issues, or monthly costs for features nobody needs.
Here at Softstorm in Bad Tölz-Wolfratshausen, we see this all the time. So before you sign up for any chatbot, ask yourself these five questions – honestly and without rushing.
Question 1: What specific problem is the chatbot supposed to solve?
"We want to do AI too" is not an answer. Is the goal to reduce the number of phone enquiries? To have the chatbot accept appointments outside of opening hours? Or to answer product questions? Without a clearly defined objective, any solution is money down the drain.
Question 2: Who are your customers – and how do they write?
An off-the-shelf SaaS solution is trained on standard German and designed for a general audience. But your customers from Miesbach or Holzkirchen ask things like: "Do you do emergency repairs on Saturdays?" or "Do you still have anything available at short notice before Christmas?" A generic chatbot often responds to these in a stilted way – or not at all – and that stands out immediately.
Question 3: Who maintains the chatbot after it goes live?
Most off-the-shelf solutions need ongoing upkeep: new FAQs, updated opening hours, new products. If nobody is responsible for this, the bot becomes outdated within weeks. So ask before you buy: how much effort does maintenance take – and is that realistic in your business?
A joinery in Wolfratshausen used a custom chatbot built on Flowise – an open-source tool for visual AI agent creation – to have 60% of enquiries outside business hours answered automatically, without any staff needing to step in. The key difference: the bot was trained directly on the business's products, pricing structures, and common customer questions. An off-the-shelf solution would not have achieved that.
Question 4: Which language model is running in the background – and where does your data go?
This is the question that almost nobody asks – and it's the most important one. Many low-cost chatbot providers route your customer data to US-based servers without you noticing. For businesses in Bavaria that need to operate in line with GDPR requirements, this can get expensive.
If you want full control over your data, you should look at Ollama: a tool that lets you run powerful language models locally – directly on your own server. No data leaving your premises, no dependency on US cloud services. For sensitive sectors such as tax advice, healthcare, or legal services in Rosenheim or Penzberg, this can be the decisive factor.
Question 5: What does the chatbot actually cost – today and in a year's time?
According to Salewski.it (2026), most AI chatbot solutions for SMEs cost between 30 and 150 euros per month. That sounds manageable – but watch out: many providers charge per conversation or cap usage. If you have a high volume of enquiries during peak season, you can suddenly end up paying three times as much. Always calculate based on the worst case, not the average.
On top of that, there are often hidden costs for integrations, support, or template customisation. If you want to connect your chatbot to your CRM, your booking software, or your website, you'll usually need an automation tool like n8n – an open-source alternative to Zapier that really comes into its own when you need to map out many custom workflows without per-user licence fees. If you don't factor this in from the start, you'll be in for an unpleasant surprise later.
Our honest take: Most AI chatbot platforms are built for American e-commerce shops – not for a plumber from Geretsried or a physiotherapy practice in Bad Tölz.
In 8 out of 10 of our projects, the first bottleneck was missing integration with existing systems – inventory management, booking calendars, trade software – not the AI itself. The real question is: how does the chatbot actually get the right information in real time?
That's a solvable problem – but not with plug-and-play. It takes a well-thought-out architecture where data sources are cleanly connected. If you want to find out more about how we approach this in our AI chatbot projects, you'll find concrete insights into our methodology there.
And one honest downside you should know about: AI chatbots clearly hit their limits when enquiries touch on legally sensitive territory – such as liability questions, warranty claims, or medical assessments. No language model in the world should be the final authority here. A good chatbot recognises these situations and reliably hands off to a human – but that has to be explicitly built in. It doesn't happen on its own.
Many of these mistakes can be avoided simply by taking a breath before you buy and asking the right questions. We're happy to help you do that – no consultant waffle, no overpriced one-size-fits-all packages.
If you want a chatbot that genuinely fits your business in Upper Bavaria – get in touch or book a free initial consultation directly. We'll take a look together at what makes sense, what it costs, and what you actually need.
We analyze your processes, identify automation potential and show you which digital systems really make sense.
