# AI for an energy supplier's contact center

> We are modernizing an energy supplier's contact center with a knowledge base and AI assistance in the client's Azure tenant. Since May 2026.

- Canonical: https://techmates.io/en/case-studies/energy-supplier-contact-center
- Client: Energy supplier contact center
- Industry: Energy
- Published: 2026-09-11
- Updated: 2026-09-15
- Services: AI agents & automation, Process mapping & optimization
- Technology: Microsoft Azure, RAG knowledge base

---

## The operator searches, the customer waits

Since May 2026 we have led the modernization of a Czech energy supplier's contact center: project and product management plus the full-stack development that brings AI into daily operations.

First we measured the starting point together with the client. Operators work across several systems at once: telephony, CRM, a knowledge base and partner portals. The knowledge base holds roughly 1,100 pages, and full-text search in it is slow and not trusted. In a six-minute call, an operator spends about 1 to 1.5 minutes searching for documents.

### We map the process before we automate it

We follow the 3A method. In Analysis we map processes and data by role and build an integration map. In Action we test each assistance scenario on a real process. In Automation we roll out assistance where operators spend the most time.

- A knowledge base with semantic search that answers with its sources
- AI assistance for operators during calls and while handling emails
- An integration layer to the existing systems
- Everything runs in the client's own Azure tenant

### Impact is measured in operations, not estimated

We track three indicators:

- Handling time
- After-call work
- First-contact resolution

The baseline comes from the measurement with the client.

## Where things stand and what comes next

The project has been running since May 2026. We lead it on the project and product side while building the knowledge base, the operator assistance and the integration layer to the client's systems. The solution lives in the client's Azure tenant so the data stays under its control.

We are testing each assistance scenario on real contact-center processes. Results will follow once we have measured them.