The Zurich Kundencenter (customer center) Lernpilot is our reference case. It's not just a sale — it's a demonstration of the shape of customer PPL Spark is built for.
Why Zurich fits
An insurance call center has the properties that make AI-assisted dialogue improvement work:
- Defined conversations. Calls have a predictable structure, a typical duration, a clear aim (resolve a customer issue).
- Measured performance. Zurich already tracks NPS. Quality change shows up in a number.
- Predictable variance. The agent varies, the customer varies — but the space of what should happen in the call is well-understood.
- Business stakes. Better conversations → higher NPS → measurable business value. Not a feel-good metric.
This is why Zurich is the "low-hanging fruit" — not because call centers are the whole market, but because they're where the shape of the problem is cleanest.
The general case
The same shape extends beyond call centers: any organization where conversations have repeatable, predictable feature sets — team meetings, negotiations, customer onboarding, difficult leadership conversations. The uniqueness of Zurich is that they already measure and already treat conversation quality as productive capacity. Most orgs don't — but the opportunity is identical.
Related opportunities in the same shape
- Other financial-services call centers — Trade Republic recently announced a shift from chatbots back to trained humans in support, suggesting the market is rediscovering conversational quality.
- Leadership teams with 360 feedback data — e.g. Convector — where cross-evaluations give us the actor data needed to model forward.
- Any B2B with a conversation-driven KPI.
External validation
Feedback from AI entrepreneurs and investors on the Zurich demo has been consistent: "You already have a viable product." The Live Companion capability demonstrated in the pilot — AI actively and intelligently intervening in live conversations — is the part that stands out most.