*This post was created with a self-made CustomGPT based on Sophie’s latest YouTube video. You can find the link to the video at the end of this article.
When we talk about artificial intelligence, many of us immediately think of chatbots, automated customer services or generative AI models such as ChatGPT. However, there are also fields of application in which data protection, system availability and security play a completely different role – for example in critical infrastructure. In this episode of Sophie’s Next AI Talk, we talk to Stefan Riesel from CREA-LOG about how AI can be put to good use there – and what requirements need to be taken into account.
Who is CREA-LOG?
CREA-LOG is a company that has specialized in the development and integration of voice dialogue systems for over 20 years – particularly in the field of voice and bot technology. Stefan Riesel has been working in sales and consulting there since the early 2000s and knows the development of the industry first-hand: from classic DTMF-based systems to today’s AI-driven bot architecture.
Critical infrastructure and AI – do they even go together?
When people talk about critical infrastructure, they are often referring to energy suppliers, water supplies or transportation systems. These areas must function at all times – failures or security gaps would have a massive impact on society and the economy. And this is precisely why particularly strict requirements apply to IT systems:
- Highest availability
- Physical and digital security measures
- Strict data protection and compliance requirements
- Limited or no cloud usage
A classic example: the grid control centers of energy suppliers. It monitors whether electricity is flowing, where outages occur and how the grid load is distributed. These systems often still run on-premise, i.e. locally on their own servers – sometimes even in physically secured rooms with barbed wire and access controls.
When the power goes out – how AI can still help
A typical scenario: The power goes out, dozens of calls land in the control center. But there are often only two or three people working there – it is impossible to answer all calls manually. The solution: A voice-based AI bot automatically asks where the call is coming from (e.g. via the zip code), checks whether a fault is known and informs the caller of the current status.
Important here:
In this scenario, the AI must not usually be cloud-based. Instead, it is installed locally as a container solution – independent of internet access, GDPR-compliant and within the customer’s secure environment.
On-premise or cloud? Hybrid solutions as a realistic option
Not every use case in the critical infrastructure needs generative AI. Sometimes classic speech recognition or a simple text-to-speech module – installed locally – is sufficient. In other cases, however, such as in the classic customer service area of an energy supplier, cloud-based AI services such as Microsoft Azure Cognitive Services can also be used.
A practical example:
An energy supplier combines a locally installed bot system with Azure AI tools in the cloud. Sensitive customer data remains local, while the AI functionality is integrated via the cloud – in compliance with data protection regulations, of course. In this way, modern customer services can be established without compromising on security.
Conclusion: Not every solution needs GPT – but every solution needs foresight
What this conversation with Stefan Riesel clearly showed us: In areas with sensitive data, it’s not always about using the most modern AI, but the most suitable. Sometimes this is local speech recognition, sometimes a hybrid combination of an on-premise system and cloud AI. It is always important to first understand the problem and then find the right technological answer.
Any further questions?
Do you have any questions? I am happy to support you, act as a sparring partner and answer your questions. I am always happy to receive your messages, preferably by WhatsApp message or e-mail.
