Utility AI has spent several years as a conference theme. In 2026, some CIS vendors are moving past the slide deck and into production. Cayenta, a division of Harris Computer, is one of them. Its AI product, Cayla (Cayenta Artificial Intelligence), is now described on the Cayenta product pages as a live capability, and a real utility deployment signed in March 2026 gives it a concrete reference point. This review covers what Cayla is, what the Lakeland deployment tells us, and where its limits are at this early stage.
What Cayla is: Agent and In-App Assistant
Cayenta describes Cayla in two distinct delivery modes, per its product pages.
The first is Cayla Agent, which manages customer inquiries across voice and digital channels. This is the outward-facing layer: a customer calls or contacts the utility through a digital channel, and Cayla Agent handles the interaction, routing and resolving without necessarily requiring a human agent.
The second is the In-App Assistant, oriented toward internal staff. Per Cayenta’s description, it can answer questions, analyze accounts, summarize operations, and execute workflows. Demonstrated actions include starting a move-in and setting up a payment plan. These are not AI generating a report for a manager to act on later. They are AI taking action inside the CIS directly on behalf of a staff user, which puts them in the agentic category.
The distinction matters for governance. An AI that answers questions carries different risk than one that executes transactions. Utilities deploying agentic workflow tools need clear policies on which actions require human confirmation and what the rollback procedure is for an incorrectly executed transaction.
The Lakeland deployment: what the record shows
The clearest public evidence of Cayla in production comes from Lakeland, Florida. The City of Lakeland Commission approved an agreement on March 2, 2026 for Lakeland Electric to implement Cayla, built in partnership with Dial AI. The agenda memo targets a production go-live around May 1, 2026, integrated with Lakeland Electric’s existing Cayenta CIS.
That makes Lakeland a first-mover, not a long-tenured reference site. At the time this post publishes, the deployment is weeks old at most. There is no multi-year performance record to evaluate yet.
What the agreement confirms is that Cayla is not vaporware. A municipality committed budget, went through a commission approval process, and signed with a specific partner (Dial AI) and a specific go-live target. For a product that only recently appeared on the Cayenta product pages, that is meaningful early validation.
Where Cayla fits inside the Cayenta CIS
The Cayenta suite is broader than the core CIS, as covered in the detailed Cayenta CIS review. It includes ServiceLink for mobile workforce, Silverblaze for customer self-service portals, SmartWorks for meter data management, and DataVoice for outage management.
Cayla sits inside this stack. Its primary advantage is that it already has authorized access to account data, billing history, service agreements, and CIS workflows. No integration project is required to give the AI access to data for a move-in or payment plan setup.
Standalone AI tools bolted onto a CIS through API connections are feasible but add cost and a surface area for synchronization errors. For a Cayenta customer, Cayla removes that tax.
Cayla Agent and the Silverblaze self-service portal address overlapping but different problems: Cayla handles inbound real-time inquiries while Silverblaze handles asynchronous self-service. The combination matters for utilities focused on contact-center deflection, a topic explored in the case for customer portals in utility services.
How Cayla compares to Oracle and SAP utility AI at a high level
Oracle’s utility AI portfolio is larger and older. Oracle Opower has a 17-year track record and over 44 million enrolled households in North America, per Oracle-reported figures. Oracle added AI call summarization to its Customer Platform in May 2025 and AI anomaly detection in Meter Data Management in June 2025. These are incremental additions to an enterprise platform with a deep installed base.
SAP Joule is the cross-application AI assistant SAP is embedding across its S/4HANA suite, including S/4HANA Utilities, targeting similar workflow acceleration but in the SAP ecosystem.
Cayla’s proposition is narrower scope and deeper CIS integration for a specific customer segment. Cayenta customers at the municipal and mid-market tier get an AI that runs inside the tools they already use, without a separate AI platform procurement. That matters for a 200,000-account municipal utility not staffed to run a large-scale AI program.
For the full CIS comparison, the Oracle vs SAP utilities analysis provides enterprise context.
Honest limits at this stage
Cayla is early. The honest assessment of any AI product with one reference site on a short deployment timeline is that the long-term operating record does not exist yet. Prospective buyers should ask:
- What is the error rate on agentic transactions (move-ins, payment plan setup), and how are exceptions handled?
- What governance controls restrict which actions the In-App Assistant can execute without human confirmation?
- What does the Dial AI partnership mean for support, model ownership, and continuity if the partnership changes?
- How does Cayla handle utility-specific edge cases: life-support accounts, delinquency restrictions, complex rate structures?
A CIS-native AI that can answer questions and execute transactions inside the system of record is a credible design for utility customer service. The open question is execution at scale over time. Utilities considering Cayla should track the Lakeland deployment’s public reporting and request updated reference sites before committing.