How Naturgy is Building Spain's Most Customer-Forward Service with HappyRobot

It's January in northern Spain, and your heating has just stopped working. Customers here don't look for support in an app — they call. And Naturgy's promise is to be there: on-site technical help anywhere in the country within three hours when a visit is needed.

Behind that phone line is now a voice AI agent — open to customers and non-customers alike — that answers the calls to Naturgy's technical service, runs a full diagnostic on each one, resolves many issues on the spot, and arranges a technician when the problem needs a professional. In its first weeks of operation, it scored 9.4 out of 10 with customers.

About Naturgy

Naturgy is one of Spain's largest gas and electricity suppliers, serving millions of residential and business customers across the country. Operating critical infrastructure, the company's promise to customers goes beyond energy and billing: free phone-based technical assistance for its appliances, and, when a problem can't be solved over the phone, a technician on site anywhere in Spain in under three hours — a commitment that matters most during cold winters in the country's northern regions.

For Naturgy, customer service isn't just a cost center. It is the product. The company's stated ambition is to be recognized as the best customer service company in Europe.

Clean Before You Move In

Most companies try to automate broken processes. Naturgy didn't. Before deploying a single AI agent, the company spent three years rebuilding the operational foundation of its client business unit from the ground up.

The transformation began in 2022, when Naturgy recognized that legacy systems and siloed data made it hard to compete with more agile, modern-stack players. Rather than patch over the problems, leadership decided to migrate the entire customer base to a new platform, starting from a blank sheet of paper. The team's guiding principle was blunt — you have to clean before you move. Running a migration on top of years of accumulated complexity is almost impossible. The work took three years, cleaning, simplifying and consolidating hundreds of products and processes, and was completed in spring 2025.

The timing proved fortunate. As the migration wrapped, generative AI was accelerating fast, and by the time the platform work closed, the team was ready to build on top of it. Because they had done the groundwork, they already knew exactly what customers were calling about, how they wanted those calls handled, and who inside the organization needed to be involved — precisely what most AI projects lack when they begin.

Voice First, By Necessity

Naturgy's customers show a clear preference for phone support. When the heating stops working in January, they don't open an app or start a chat thread — they call, and they expect to reach someone who can solve their problem. Internally, the team had already built voice-agent prototypes as a learning exercise — not to ship, but to understand the technical challenges firsthand, so that conversations with vendors could be grounded and honest.

Armed with that internal knowledge, Naturgy evaluated close to 30 vendors, from hyperscalers to boutique startups, and ran structured proof-of-concept trials with the most promising: HappyRobot. The selection criteria were clear: naturalistic conversation quality, production-grade reliability, and a team willing to co-build, not just deploy.

Thirty Vendors. One Demo. Three Months.

The POC with HappyRobot focused on Naturgy's most common — and most consequential — customer-service moments: invoice explanation, meter-reading submission, FAQ handling, and technical troubleshooting. A mock environment replicating Naturgy's systems was built so the agent could demonstrate reading from and writing to real-style data without touching production.

Shown internally to a 100-person “friends and family” group, the response was immediate and enthusiastic. The POC went from kickoff to delivery in three months, and HappyRobot was chosen to power Naturgy's primary technical-service line — the highest-volume, highest-stakes channel.

This Isn't Call Deflection

What makes this deployment significant is not just that a voice agent answers calls — it's the depth of what it does within each one. The line is open to everyone, customers and non-customers alike, and the agent runs a full diagnostic on every call before any question of contract or coverage comes up. This is a full-service operational agent embedded in Naturgy's live systems, executing a multi-step workflow on every interaction:

• Full issue triage and on-call resolution: the agent runs a structured diagnostic and walks every caller — customer or not — through troubleshooting live. In many cases the issue is resolved entirely on the call, with no technician required.
• Identity verification via OTP: when account-specific information is needed, the agent authenticates the caller with a one-time passcode sent to the phone number on file, meeting Naturgy's security requirements without slowing the conversation.
• Contract and coverage lookup: the agent pulls the customer's contract in real time and checks whether the reported issue falls under their agreement, communicating clearly when coverage doesn't apply.
• Technician dispatch when needed: if the issue can't be resolved on the phone, a technician is sent to the customer's home. For customers with a maintenance contract this is covered under their agreement; for non-customers, a technician can still be dispatched at a fixed, up-front price.
• CRM write-back with full classification: after every call the agent writes a detailed, structured summary back to the CRM — issue type, triage outcome, coverage status, resolution path — with no manual logging and nothing lost in handoff.
• Anomaly detection and alerting: the system monitors call-volume patterns by outcome in real time, flagging deviations before they surface as operational problems.

In line with Naturgy's commitment to transparency, customers are told they are being assisted by an AI agent and can ask to be transferred to a human agent at any point in the conversation. The assistant is built to answer 100% of incoming calls simultaneously, with no wait times, and Naturgy is expanding its scope deliberately — validating quality at each threshold before adding volume. The service has been live only a matter of weeks, and the company treats the early results as a starting point, not a finish line.

In Customers' Own Words

Real end-of-call feedback, collected via post-call survey:

“They answered quickly, told me they'd help me, and that a technician would call me back. Everything went smoothly.” — 10/10

“The care and attention they showed me — so thorough and so kind.” — 10/10

Always On

One of the clearest results of the deployment is its consistency. Many customers call when something at home has broken — often stressed, sometimes needing more time to explain the problem or a step repeated. Whatever the hour, they get the same calm, patient, methodical interaction.

Just as important, this doesn't replace Naturgy's people — it repositions them. By taking on the most repetitive, high-volume interactions, the agent frees customer-service professionals for the cases that call for judgment and empathy.

“Our AI agent handles the most routine tasks with precision and consistency, which frees our teams to focus on the situations that require judgment, empathy and a more personal relationship with the customer.”

— Terrell Sheridan, Head of GenAI Transformation, Naturgy

Part of the Team

This matters in a European enterprise context. Naturgy is a large, complex organization — change management is real, stakeholder alignment is hard, and any solution that can't navigate that reality will stall regardless of technical quality. HappyRobot's model is to embed its forward-deployed engineers directly in the client's operations.

“We work as an extension of Naturgy's own team — in their offices, talking directly to their people. In a project like this, being embedded isn't a nice-to-have; it's what turns a promising demo into a solution that actually runs in production.”

— Antonio González Ferreira, Forward Deployed Engineer, HappyRobot 

For Naturgy, that combination of deep technical capability and European implementation sensibility is what set HappyRobot apart from vendors with comparable technology but greater cultural distance.

What Naturgy Is Building Toward

Naturgy's ambition isn't cost reduction. It's transformation. The company is actively exploring new use cases to extend the technology across its operations. The mindset behind it is less about today's numbers than about their trajectory: success is measured not by where a KPI sits right now, but by whether it will keep improving.

It is also building toward a model where human and AI agents co-manage service — not as a replacement dynamic, but as a hybrid workforce.

“Our teams will manage both AI and human agents. That's not something we're afraid of — it's the direction we're building toward.”

— Terrell Sheridan, Head of GenAI Transformation, Naturgy

What Naturgy and HappyRobot have built isn't a pilot. It's the start of an operating model — designed not around what AI can do today, but around where customer service in Europe is heading.

‍