<!-- Markdown version of https://www.infosistema.com/industries/utilities/ · last reviewed 2026-10-20 -->

# AI agents for energy, water and telecom utilities

A utility answers millions of customers through billing, metering and field systems that were never designed to talk to each other. AI agents can carry the requests and exceptions between those systems. Infosistema connects them, keeping them away from the controls that run the network itself.

## At a glance

- Infosistema connects AI agents to the customer, billing, metering, asset and field-service systems of energy, water and telecom utilities.
- Typical first processes are customer service requests, billing and meter-reading exceptions, field work-order preparation and supplier contract renewals.
- Agents work on commercial and back-office processes; they don't act as safety components of the grid or the water network, which the EU AI Act treats as high-risk.
- Every engagement starts with a three-week analysis sprint that ends in a decision per process, and every agent starts in shadow or supervised mode.
- Eneco, the Dutch energy company, uses Infosistema's DMM in its OutSystems projects, and Infosistema's clients include NOS, Serviços Municipalizados de Abrantes and ERSAR, the Portuguese water and waste regulator.

[Talk to us](https://www.infosistema.com/contact/?start=agentic-ai&industry=utilities)

## Where do AI agents fit in a utility?

Between the customer and the operational systems, where a request touches billing, metering and field service before anyone can answer it. In most utilities, the same 4 processes come up first.

### Customer service requests

The agent handles moves, contract changes, payment plans and outage questions end to end: it reads the request, checks the account, updates the systems involved and replies, with a person on anything unusual.

- **Stays with people:** disconnections, vulnerable-customer cases and every complaint that needs a judgment call.
- **Systems:** CRM, billing, customer portal, contact-centre platform.

### Billing and meter-reading exceptions

The agent investigates the bills that fail validation: estimated reads, sudden jumps, missing data from smart meters. It proposes a correction with the evidence and sends the clear ones back to the billing run.

- **Stays with people:** credits above the agreed threshold and any change to a disputed account.
- **Systems:** billing engine, meter data management, CRM.

### Field work-order preparation

Before a technician goes out, the agent assembles what the job needs: asset history, permits, previous visits, customer access notes. It also flags the orders that can be solved remotely.

- **Stays with people:** scheduling decisions and everything that happens on site.
- **Systems:** work-order and asset management, GIS, CRM.

### Supplier and service contracts

The agent finds supplier, maintenance and software contracts across inboxes and drives, extracts the terms and warns before each renewal, so a contract renews because someone chose it.

- **Stays with people:** keep, renegotiate or cancel.
- **Systems:** email, Google Drive, ERP; [bizSupply](https://www.infosistema.com/products/bizsupply/) does this as a product.

## What makes agents hard to run in a utility?

Critical infrastructure rules and systems that each hold one piece of the customer.

- **EU AI Act.** AI used as a safety component in the supply of electricity, gas, heating or water is high-risk. We keep agents on the commercial and back-office side and write that boundary into each agent's scope.
- **NIS 2.** Energy, water and telecom operators are essential entities. Agents and their providers fall inside your cybersecurity and supply-chain obligations, with logging and incident reporting to match.
- **Regulated customer processes.** Disconnections, social tariffs and complaint deadlines are set by the regulator. These steps keep a person in the loop.
- **Fragmented data.** Customer, meter and asset data rarely agree. Most of the readiness work in the sprint is about data, before any agent is built.

## How does Infosistema start?

With a three-week analysis sprint on a defined scope, for example residential customer requests or billing exceptions in one region.

1. **Week 1:** map the processes and systems, and short-list up to five candidates.
2. **Week 2:** score each process for impact and for readiness, and analyse it step by step against a baseline measurement.
3. **Week 3:** estimate build and monthly operation for each agent, then recommend one of four outcomes: agent, agent with conditions, classic automation, or not yet.

Quick wins can start in supervised mode from week 2. Each agent then earns more autonomy only when its results against the baseline justify it, with your approval at a monthly committee.

## Which Infosistema products and partners help?

- **[bizSupply](https://www.infosistema.com/products/bizsupply/):** finds supplier contracts in your mailbox and Google Drive, extracts the supplier, dates, value and renewal terms, and alerts 90, 60, 30 and 7 days before a renewal.
- **[DMM](https://www.infosistema.com/products/dmm/)** and **[DMM Infinity](https://www.infosistema.com/products/dmm-infinity/):** data migration and test data for the OutSystems and Mendix applications many utilities run their customer and field processes on.
- **[Google CoE](https://www.infosistema.com/center-of-excellence/google/data/):** BigQuery and Looker for meter, billing and asset data, and Gemini Enterprise Agent Platform for agents on Google Cloud.
- **[Vesster](https://www.infosistema.com/partners/vesster/):** governed agents with graduated autonomy and an audit trail, deployable on-premises.

## What has Infosistema built for utilities?

- [Eneco delivers its project with DMM](https://www.infosistema.com/case-studies/eneco-energy-platform/): data migration in every stage, from testing to production. "It is almost mandatory to use DMM for the OutSystems projects you're working on," says Marc Sibeijn, Project Manager at Eneco.
- Utility and regulator clients on our [clients page](https://www.infosistema.com/clients/) include NOS, Serviços Municipalizados de Abrantes and ERSAR.

## Frequently asked questions

### Can AI agents run grid or network operations?

We don't put agents in control of the grid or the water network. Under the EU AI Act those are high-risk safety uses, and they belong with the operational technology vendors. Our agents work on customer, billing, field-preparation and procurement processes.

### Which utility processes are best for a first AI agent?

Customer service requests and billing exceptions: high volume, clear rules and data that already sits in your systems. The analysis sprint checks this for your case and says where classic automation is cheaper.

### How do agents fit with NIS 2?

They're treated as part of your ICT estate: least-privilege access, logging of every action, an owner per agent and an agreed shutdown procedure. Infosistema works under ISO/IEC 27001 certified practices.

### Will customers know they're dealing with an AI agent?

Yes. The EU AI Act requires people to be told when they interact with an AI system, and we design the replies and channels that way from the start.

### Do we need to replace our billing or CRM system?

No. Agents work through the APIs of the systems you have, and Infosistema builds a service layer where there isn't one.

## Let's put AI to work in your utility

Tell us which process you'd like to look at first, and we'll reply within 1 business day.

[Talk to us](https://www.infosistema.com/contact/?start=agentic-ai&industry=utilities)
