Intelligence · Phase 3
Process Mining
Real process flows from the event log. Where are the bottlenecks, where the loops, where does reality differ from the drawn-out target process?
What it does
01
Discovery
Process-Mining-Pipeline reconstructs the actual process model from events — no workshops.
02
Conformance check
Target vs. actual. Where does practice diverge from the plan and why?
03
Bottleneck analysis
Which step causes the longest wait. With an improvement suggestion.
How it runs
- 01
Read event log
Daily batch + on-demand. The AI layer works asynchronously on the log.
- 02
Build map
Process-Mining-Pipeline builds flow maps per process family.
- 03
LLM explains
An LLM condenses to natural language: "Order 247 took 4 days — avg 1.5. Bottleneck at dispatch."
What you get
- Process maps as interactive visualisations
- Top-5 bottlenecks per week
- Conformance score per process type
Requires
- → AI layer + ≥ 6 weeks of event history.
In the example
Daily
batch + on-demand
≥ 12
recommendations / year (sample)