80% of Exception Processing Automated. 20 Hours a Week Back.
A utility billing company was drowning in manual meter exception processing across 20,000 units. Vertice Labs built a custom automation agent that classifies, resolves, and generates work orders automatically.
A Family Business Capped by Manual Work
IMR Billing is a family-owned utility billing services company. They process meter readings, calculate usages, and send out utility bills for properties across their portfolio. Isabel Mingle, the owner, manages the day-to-day operations, including the billing cycles that keep the business running.
At 20,000 meters on a monthly billing cycle, the volume was significant. But the real problem was not the volume itself. It was what happened when meter readings came in wrong.
Isabel and her husband had the same conversation repeatedly: there has to be a better way to do this. The breaking point was not a single event. It was the accumulation of grinding through the same process, property after property, month after month, knowing the business could not grow if everything depended on one person doing manual work.
When Isabel tried to train someone else, the math and decision logic were difficult to explain. The expertise lived in her head, not in a system. The business was capped by how many exceptions one person could process manually.
Every billing cycle, meter readings arrive with exceptions. Each one requires investigation, classification, and resolution — a process that consumed entire billing cycles.
300 Exceptions per Property
A single property with 500 units could produce 300 exceptions. Zero reads, negative reads, abnormally high usage, stuck meters. Processing those took Isabel two to three hours of manual math, cross-referencing between two separate systems that do not talk to each other.
Weekends and 18-Hour Days
When billing cycles were heavy, Isabel worked weekends and 16- to 18-hour days to hit deadlines. The work was exhausting and repetitive. And the worst part: it was hard to hand off. The expertise required years of domain knowledge that could not be easily transferred.
Meters processed monthly — with two separate systems that do not talk to each other. Every exception required pulling data from one system, cross-referencing in another, and applying years of domain expertise to classify and resolve.
Vertice Labs embedded with IMR Billing as a development partner, not a vendor. The engagement started with understanding exactly how Isabel processed exceptions — what the classification logic looked like and where the pain was concentrated.
Exception Classification
The system identifies and categorizes each exception type — zero reads, negatives, high usage, stuck meters — using the same logic Isabel had been applying manually for years. These were known patterns, but they had never been codified.
Automated Resolution
Once classified, the system applies resolution logic automatically based on historical patterns and Isabel’s domain rules. The same decisions she made hundreds of times a week, now running in seconds.
Work Order Recommendations
When a meter shows repeated exceptions, the system flags it for physical repair and generates a work order — something that previously required Isabel to track problems on a notepad and manually schedule service.
Adapting Around Third-Party Constraints
The build depended on a third-party meter data API that was not always ready. Rather than stalling, the team adapted — reprioritizing features and working around API limitations to keep momentum.
The IMR Agent: Automated Exception Processing
A custom automation platform that handles the full exception processing pipeline: ingestion, classification, resolution, and work order generation. It integrates with IMR Billing’s existing meter data provider and billing software, processing a full data cycle in approximately 36 minutes.
Automated Classification
Identifies and categorizes every exception type using codified domain rules built from Isabel’s years of expertise.
Intelligent Resolution
Applies resolution logic automatically based on historical patterns. What took hours now runs in minutes.
Work Order Generation
Flags meters with repeated exceptions and generates work orders automatically — 99 generated in a single run.
Systems Integration
Connects two previously siloed systems — the billing software and the meter reading platform — into a single automated pipeline.
Full Ownership
Documentation, architecture diagrams, and classification logic all handed over. No vendor lock-in. No dependency on Vertice Labs to keep the lights on.
Product Strategy
Vertice Labs worked with Isabel on go-to-market thinking — use the platform internally first, let it mature through real usage, then take it to market as a hardened product.
There are companies in this space with 40 people doing this same manual processing. The market for this kind of automation is real. Isabel is already thinking about Phase 2: integrating leak detection to extend the platform further.
Weekend and overtime work during billing cycles eliminated. Isabel got her time back — the most important thing.
Full data ingestion runs in 36 minutes. A cycle of 10 properties that would have taken an hour processed in five minutes.
Training new team members is now possible — the domain logic lives in the system, not in one person’s head.
Phase 2 already in planning — leak detection integration to extend the platform further and open new revenue opportunities.
In Isabel's Words
“The agent has given me self-confidence, has saved time for my business, and it has allowed me to dream bigger for my company.”
“We found that when we found you, the process was easy. It made us breathe better, knowing that we’re going to get to where we need to be.”
“The collaboration between Vertice Labs and IMR was seamless. That’s the right word.”
“I was like, oh my God, this is it. This is what we need. It’s happening. I had my timer on. What would have taken me an hour was done in five minutes.”
“These are ideas that we’ve had in the past year that we know are not out there yet. Something that is coming out of my brain is becoming a reality.”
Ready to build something like this?
20 minutes. We'll figure out if this is the right fit.