AI consulting & development
Does this really need AI?
Much of what is called AI today is algorithms, statistics or automation. First we find out what your task needs. Then we build exactly that.
Discuss your project →Not every task needs a neural network.
A rule is cheaper, faster and easier to check. So we first look at whether algorithms or statistics are enough.
Models pay off where rules stop working. There we build chat agents, classifiers and process automation based on AI.
The first check
Four questions before the first line of code
Four example tasks
- Can you write the rule down? No Yes Rule in code No model needed
- Are past figures enough? No Yes Statistics Forecast from your data
- Are there enough real examples? No Yes Classifier Model learns from examples
- Is it about free language or documents? No Yes LLM or agent Language model with your data
A simple if-then rule is enough.
Simplified view. In practice many solutions combine several stages.
When it does need AI
What we build
Chat agents
Assistants answer questions from your documents and carry out tasks in your systems.
Question Answer with source
Classifiers
Models assign requests, documents or cases to a category.
Ticket Category
Document extraction
Contracts, invoices and reports become structured data.
PDF Data fields
AI-based automation
The process stays normal automation. The model only takes the steps that rules cannot handle.
Email Case in the system
MCP servers
AI applications access your systems through a controlled interface.
AI application Your system
Data analysis and forecasts
Statistics and models on your data, from reports to predictions.
Raw data Forecast
AI is often one part of a larger workflow. We capture and improve that workflow first. To process digitalization →
AI consulting
Where does AI pay off for you?
We go through your workflows with you and test each idea with the four questions. If an idea does not need AI, we tell you.
What you have afterwards Example
| Use case and benefit | Solution | Data | EU AI Act | Next step |
|---|---|---|---|---|
| Route tickets to the right department | Classifier | Examples available | Minimal risk | Build prototype |
| Answer questions on internal policies | LLM | Clarify access | Limited risk | Build prototype |
| Send invoices from € 5,000 for approval | Rule | Available | No AI | Implement directly |
| Pre-sort job applications | Classifier | Personal data | High risk | Put on hold |
The EU AI Act view is a first assessment and not legal advice.
Approach
From use case to operation
- Focus
Check the task
We find out whether the task needs AI and which data exists.
Prototype with real data
A small prototype shows early how good the result gets with your data.
Build it into the process
The solution becomes part of your systems, with permissions, logging and human control.
Run and measure
We measure quality and cost in operation and keep improving the solution.
FAQ
Common questions on AI projects
How do we know whether we really need AI?
What is the result of AI consulting?
Does our data have to go to the US?
How accurate are such systems?
What does the EU AI Act mean for our project?
Technologies
What we build AI solutions with
Other services
Our office is at Glacisstraße 11 in Graz. We work for companies across Austria. Software development in Graz →
Every project starts with a conversation.