Develop intelligent solutions.
AI and Operations Research where it makes sense. From data analysis to automated decisions.
From data to decisions.
Analytics is not a single step but a ladder. The higher the rung, the more the system takes over and the greater the operational leverage.
What happened?
Make data visible. Reports and dashboards show what was.
Why did it happen?
Understand causes. Patterns and correlations explain the why.
What will happen?
Anticipate the future. ML models estimate demand, failures or risks.
What should I do?
Derive decisions. Optimization models recommend the best action.
Which rung fits your problem?
A well-automated process with standard interfaces is often worth more in practice than a complex model without integration.
Not every problem needs an AI model. What matters is where in the chain today the biggest bottleneck sits.
Who should make the decision?
Both approaches have their place. The choice depends on stakes, speed, trust and how many implicit constraints your decision carries in practice.
Human in the Loop
Human as decision-makerUseful when stakes are high, regulatory requirements exist or when context is missing. Human capacity limits throughput.
AI Agent
System as decision-makerThe agent decides, acts and self-corrects. Human feedback is requested selectively, not as a standard step. The bottleneck becomes system quality and trust.
When does Human in the Loop make sense?
Whenever decisions have high stakes, need to be regulatory-compliant, or when the model alone lacks enough context. A model recommends, a human reviews and decides. This is not a step backward; it is often the right architecture.
When is an AI agent worthwhile?
When decisions are repetitive, time-critical or occur in high volume, and the consequences of a mistake are manageable or correctable. Typical cases: replanning on disruptions, automatic resource allocation, reactive adjustments in live operations.
Do I always have to choose?
No. Many systems combine both: routine decisions run autonomously, outliers or edge cases escalate to a human. The threshold for escalation is itself a design decision and can be data-driven optimized.
What are implicit constraints and why are they so difficult?
Explicit constraints are documented: capacity limits, qualification requirements, legal rules. Implicit constraints live in the minds of specialists: this supplier is never used for urgent orders; this shift nobody does twice in a row; we never push this machine to its limit. They appear in no manual, but an algorithm that ignores them produces plans that nobody follows in practice. Making them visible, questioning them and consciously including or excluding them is often the most demanding work in the entire project.
Live Demo
Simplified example app: Shift Planner
An interactive short example of what such a solution can look like as a web application. Qualifications, availabilities and shifts are freely configurable; the algorithm calculates the plan in seconds. The example is intentionally stylized; real solutions adapt to more complex requirements.
| Monday | Tuesday | Wednesday | Thursday | Friday | |
|---|---|---|---|---|---|
| Early | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD |
| Late | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD |
| Night | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD | ITS NA PÄD |
What is your use case?
Every planning problem is different. Tell us about yours and we will assess together which approach fits and what can realistically be implemented.
Implemented solutions
Selected use cases from healthcare, manufacturing and services, as reference and starting point for your own application.
Nurse Tours
A nurse calls in sick at short notice. The system automatically redistributes their tours, respects qualification requirements and minimizes detours. The service runs without interruption.
- Automated tour assignment
- Qualifications and time windows considered
- Route optimization in real time
- Short-notice adjustments on absences
Car Sequencing
What happens when parts are suddenly missing? The system automatically shifts affected vehicles, fills the gap and respects blocking rules. The line keeps running.
- Optimized sequencing to final assembly
- Blocking rules and inventory control
- Short-notice replanning on disruptions
Master Production Scheduling
A supplier reports a delay. The system reprioritizes critical orders, reallocates capacities and immediately shows which production weeks are affected.
- Assign orders to production weeks
- Align capacity and volume
- Prioritize critical orders
- Planning under bottlenecks
No matching solution here?
Many planning problems are structurally similar. Let us check together whether an approach can be adapted to your use case.