Steve Dawson is the founder and CEO at POWERCONNECT.AI.
For decades, electric, water and gas utilities have approached Customer Information System (CIS) modernization with a familiar strategy: When technology becomes outdated, replace it.
That decision can initiate lengthy procurement processes, complex data migrations, investments and years of implementation. Yet artificial intelligence (AI) offers another possibility: improving how employees and customers interact with enterprise systems without necessarily replacing the platforms responsible for billing millions of customers.
The next generation of CIS modernization may focus less on replacement and more on unlocking the value already embedded in utility technology.
The Real Challenge May Not Be The CIS
Many utilities operate CIS platforms that have served them reliably for decades. These systems manage customer accounts, billing, payments, meter readings and service orders. They also contain rate structures, business rules and integrations that would be costly to replicate.
Nevertheless, customer service representatives often navigate multiple screens and applications to answer a question. The billing engine may work correctly while its user experience falls short of modern expectations. That distinction matters. Does a utility need an entirely new CIS or a better way to interact with the one it already has?
APIs And MCP: Building The Connection Layer
Application programming interfaces (APIs) let applications exchange information and perform authorized operations. A modern user interface can use APIs to retrieve billing histories, account balances, payment records, meter data and service requests, while the existing CIS remains the system of record.
The Model Context Protocol (MCP) adds another capability. This open standard enables AI applications to connect with external tools and data sources through a consistent interface. APIs supply underlying system functionality; MCP can expose approved capabilities to AI assistants and agents.
For example, a utility might offer tools for checking an account balance, reviewing consumption or retrieving service order status. An AI copilot could use those tools, subject to the employee’s permissions, without requiring a custom connection for every new AI experience.
MCP is not a replacement for APIs, and neither guarantees interoperability or security by itself. Utilities still need integration architecture, authentication, authorization, monitoring and clear ownership of each transaction.
AI Copilots Change The Employee Experience
Imagine a customer calling because a service transfer has not been completed. Today, a representative may need to open the CIS, locate a service order, check field activity and contact another department.
With a copilot embedded in a redesigned interface, the representative could ask: “What’s delaying this customer’s service transfer?”
The copilot could gather authorized information from connected systems, identify an outstanding inspection or pending work order and present the next step for review. It could also explain which source supports its answer, making the response easier to verify.
This approach can support billing inquiries, payment questions and high-consumption investigations. The objective is not to let AI invent answers or bypass established workflows. It is to reduce navigation and give employees useful context when needed.
From Copilots To Agentic Workflows
Copilots generally help people interpret information and prepare actions. Agentic AI can go further by coordinating steps across systems. A governed agent might collect information needed for a service request, validate required fields, prepare an update and route it for approval. Connected through APIs or MCP tools, it could help bridge billing, payment and field operations applications.
But greater capability brings greater responsibility. High-impact actions should require approvals, with safeguards against incorrect data, unauthorized requests and unexpected behavior. Human oversight must match the operational consequences of the task.
Security Starts With Architecture
Utilities handle sensitive customer information and operate critical services. Introducing AI should not mean copying entire customer databases into another platform.
Where architecture allows, an intelligent interface can request only the information necessary for an authorized task, leaving core records within existing systems. This may reduce duplication, although temporary processing, logging, vendors and model-provider retention policies still need careful review.
Security controls should include encryption, role-based access, audit trails, tool permissions, prompt-injection defenses and transaction validation. The same employee who cannot change an account in the CIS should not gain that power simply by asking an AI assistant.
Utility-Specific Intelligence Is Essential
A general-purpose chatbot may write fluent answers without understanding utility operations. Accurate billing explanations can depend on rate schedules, meter intervals, estimated reads, seasonal usage and local policies.
That is why AI must be grounded in authorized source data and utility-specific rules. Responses should distinguish verified facts from suggestions, and representatives should be able to inspect the evidence behind recommendations.
A polished answer is not necessarily a correct answer. For utilities, trust depends on traceability and operational accuracy.
Modernization Should Begin Before The RFP
Timing also matters. By the time an organization releases an RFP for a new CIS, leadership may already have determined that replacement is the answer. Alternative approaches can be harder to consider once requirements and budgets reflect that assumption.
Utilities should carefully assess API readiness, integration constraints, user experience, security and AI opportunities before deciding whether replacement is necessary. Some aging platforms genuinely lack vendor support, essential functionality or a viable security posture. Those systems may need replacement.
Others may support incremental improvements that preserve billing logic, minimize migration risk and provide value sooner. A disciplined assessment should compare both paths against outcomes, cost and maintainability.
An Intelligent Experience Layer
The traditional CIS will likely remain central to accurate billing, financial transactions and customer records. What changes is the way people access those capabilities. Modern interfaces can simplify everyday work. APIs can connect systems. MCP can standardize access to approved AI tools. Copilots can explain information, while agentic workflows can coordinate controlled tasks.
Success should be measured by resolution times, accuracy, customer satisfaction, employee productivity and reliability, not by how much software gets replaced. Before asking, “When should we replace our CIS?” Utility leaders should ask, “What could we improve with systems we own?”
Modernization doesn’t begin not with replacement but rather with a more intelligent way to use what is already there.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?