Building an AI Agent to Simplify Grid Connection Applications at Elektro Ljubljana | Comland x FlowX.AI
Elektro Ljubljana receives ~5,000 grid-connection requests a year, and most arrive incomplete or incorrect. Comland and FlowX.AI are building an AI agent that fills the form from a photo of the customer's electricity bill, simplifying the application process.
You want to charge an electric car at home, or put solar panels on the roof. Before any of that, you need approval from your electricity distributor. In Slovenia this is done through the enotna vloga, the unified grid-connection application used by all electricity distribution companies.
The form asks you to know technical details customers may not readily recognize: your metering point, your connection capacity, how your building is classified. The electricity bill already contains information a utility needs. Yet, completing an application can still mean finding that information, interpreting unfamiliar fields and typing it all in again.
Elektro Ljubljana receives around 5,000 of these a year, most of which contain errors. With Comland, we're building an AI agent that closes the gap: upload a photo of the bill, the agent reads it, fills in what it can, and asks for the remaining information in plain Slovenian.
The use case is in development and moving towards a pilot. The goal is clear: help the request arrive complete the first time, with less administrative work for customers and the teams receiving it.
The workflow concerns roughly 5,000 requests a year, submitted through the enotna vloga. Customers must understand the utility's terminology and translate their circumstances into the required fields. An incomplete submission can create another round of questions, corrections and waiting before substantive processing can begin. Staff must review the application, request corrections and wait for the customer to respond. This creates delays for customers and repeated administrative work for the utility's teams.
The operational opportunity is to reduce that avoidable work on both sides. A simpler application journey matters because it can improve the quality of what enters the process.
The planned customer journey brings together three steps:
This shifts the burden of interpreting the form towards the service itself. Customers can provide information in a more natural way, while the utility aims to receive a more complete application. The AI-assisted journey aims to make a complex utility application easier for customers to complete.
A useful interaction must also fit the organisation behind it. Comland is building the implementation on the FlowX.AI platform: Comland brings the utility-sector expertise and the form and field layer, while FlowX.AI is the underlying AI platform. The interesting part is that this agent does institutional work, not individual work - because in a regulated business, "the AI did it" is not an audit trail.
The work extends beyond reading a document to the requirements around the agent: integration with existing portals and email, consistent behaviour, personal-data handling while staying compliant with GDPR, explainability and an audit trail.
These requirements turn a convenient interaction into a workflow the institution can stand behind. For example, unclear information needs a defined path to clarification, and staff need to understand how an application was populated. The pilot will be the place to validate the experience and the controls together.
As the build moves towards pilot, three useful measures are (1) requests completed on first submission, (2) the follow-up contacts needed per request, and (3) the time until an application is ready for processing. These would connect the customer experience to operational value. The purpose is to establish whether better intake reduces rework and helps staff begin processing sooner. Any efficiency gains remain to be validated.
For other regulated organisations, the practical insight is to start where information enters a workflow. Improving that first handoff can reduce the work that follows.
Tanja Blatnik of FlowX.AI and Rok Kreslin of Comland will discuss this project in their joint keynote, "AI in the Core: The Biggest Opportunity in Regulated Business". Learn more about bringing AI into your own business workflows at www.comland.com and www.flowx.ai
