Overview

Modernizing customer service in high-demand utility environments

Utility providers manage millions of customer interactions every year. Billing inquiries, account status requests, reconnection processes, and administrative questions drive significant operational costs and increase wait times.

To tackle this challenge, one of Argentina’s leading natural gas distributors —handling over 87,000 monthly customers and averaging 3,971 daily interactions— embarked on a digital transformation journey. The goal was clear: migrate repetitive interactions to intelligent digital channels, cut traditional call-center costs, and address real customer friction without compromising support quality.

Together with Santex, the organization conducted an extensive Discovery and Business Analysis process, leveraging Speech Analytics and Natural Language Processing (NLP) over a dataset of 1,200+ actual support calls. This insight enabled the design of an AI-driven conversational strategy, UX optimization, and a dedicated Middleware architecture focused on operational efficiency and a measurable Return on Investment (ROI).

The Challenge

When support demand outgrows traditional legacy systems

The organization faced a critical operational bottleneck: 46.4% of customer interactions were concentrated in the phone channel, the business's most expensive support avenue. Deep-dive root-cause analysis revealed that the contact center was frequently acting as a "help desk" to handle failures occurring within digital and logistical processes.

Support demand was heavily concentrated around three critical friction points:

• Cuts and reconnections: high-urgency, high-frustration calls triggered by failed technical field visits or post-payment anxiety.

• Debt management and reverse collections: customers with a willingness to pay ended up calling support because paper bills were delayed or digital platforms had usability barriers.

• Frustrated omnichannel journeys: nearly 4 out of 10 phone calls came from users who had already tried to self-manage through the website and failed.

Compounding this was a legacy technology challenge: the company’s primary SAP interface used an asset-centric data model rather than a customer-centric one. This imposed a heavy cognitive burden on agents, forcing them to manually mine data across multiple tabs to reconstruct each customer’s history. The goal was not simply deploying a basic chatbot; it was redesigning the end-to-end customer experience.

The Solution

Conversational AI built to resolve, escalate, and drive measurable outcomes

Santex designed and deployed an AI Core combined with an intelligent middleware layer that seamlessly integrated with the utility’s existing web services, avoiding the need for a costly, high-risk overhaul of the central ERP system.

The solution upgraded a traditional informational bot into a resolutive conversational assistant with the following capabilities:

• Automated first-mile identification: recognizing and validating accounts, identifying the contact user, and updating contact information from the first interaction.

• Guided high-frequency workflows: automation for debt checking, duplicate bill generation, and real-time reconnection order tracking.

• Smart escalation with human fallback: natural language processing detects frustration or explicit help requests, transferring the full conversation context to live agents through Wise CX.

• Agent optimization tools: a level-zero 360° customer view consolidates vital data, speeds up handling times, and uses AI to categorize closing reasons automatically.

The Approach

A phased rollout strategy focused on adoption and continuous improvement

To balance modern innovation with the non-negotiable operational continuity of an essential utility service, Santex structured an agile roadmap aimed at maximizing time-to-value under a clear philosophy: first function, then understand, next automate, and finally consolidate.

1. Foundations and integration: establishing real-time chat data flow by embedding the AI Core into existing digital channels, including WhatsApp and web bot, via Wise CX.

2. Identity protocol and intelligence: activating NLP models to interpret intent, understand complex contexts, and securely validate customer identity.

3. Business workflows and autonomy: deploying back-office AI agents to process transaction-heavy happy-path requests autonomously.

4. Consolidation and MVP stabilization: delivering a robust, secure, and repeatable MVP ready for IT, business, and data security review.

Business Impact

Tangible ROI and enhanced operational efficiency

By tying product strategy and artificial intelligence into user-centered design, the initiative unlocked a highly efficient, scalable service model backed by a rigorous KPI and OKR framework:

• 71.46% effective annualized IRR projected through comprehensive financial ROI modeling.

• Estimated payback period of 2.25 years, generating an accumulated ROI of 204.6% over the evaluated horizon.

• Structural call center cost reductions by shifting thousands of repetitive inquiries to digital self-service available 24/7.

• Emergency channel decompression through automated reconnection status updates that kept administrative inquiries away from critical emergency lines.

• Optimized average handling time by removing systemic data over-validation and pre-loading context for live agents during digital handoffs.

This project shows that driving major efficiency through generative AI does not require tearing down core infrastructure. It requires deploying intelligence strategically where customer pain and operational cost intersect.