Comcast Technology Solutions case study

At Comcast Technology Solutions, billing is anything but one-size-fits-all. Seven or eight analysts each support their own product area, with different commercial models, workflows, and operational requirements. In ad delivery, even a seemingly simple billing task can involve hundreds of invoices, complex workflow logic, and business rules that are difficult to capture in a single playbook.

As complexity kept increasing, Chris Skalet, senior manager of finance at CTS, knew he needed to find a way the team could spend less time executing repetitive tasks and more time helping the business move forward. His answer: put Zuora AI to work inside the billing environment, not as a replacement for finance expertise, but as a way to extend it.

Starting small, on purpose

Chris first discovered Zuora AI while working in the tenant. He tested it in a sandbox, then tried it against a small sample of real production data he already understood well. He didn’t start with a grand transformation project, but by looking for tasks that were clear, repetitive, and tedious enough for an AI assistant to help with.

One of the first was an audit task: cross-checking communication profiles against account types to catch setup mistakes. It’s the kind of thing that’s easy to get wrong once and never notice, especially across a distributed team. The AI surfaced a handful of accounts with the wrong profile attached and Chris’s team fixed them. Small, but it told him the tool could actually see what he needed it to see.

Automating the manual click-through

From there, they decided to tackle a much bigger task. CTS’s system generates draft invoices for certain customers, and those invoices still need to be posted. Most customers were already covered by automated bill runs, but in ad delivery, some accounts ran through workflow instead, generating dozens or hundreds of invoices in a single monthly cycle.

The process was simple, but also simply exhausting. An analyst, or Chris himself, clicked through the invoices one at a time. At close, he recalls posting roughly 60 invoices for a single customer. “It was kind of a mind drain,” Chris recalls. “I could be doing something better than that with my time.”

With Zuora AI, Chris realized he could hand over the invoice list and instruct it to post them. Nothing changed about who decided what should happen; the AI didn’t act on an inferred intention, only on his explicit instruction. What disappeared was the manual sequence of clicks. The value wasn’t only the minutes saved on one batch; it was removing a mindless, error-prone task from the team’s operating rhythm and handing analysts back the time for work that actually required judgment.

Fixing workflow logic at the speed of an idea

An even bigger opportunity turned out to be workflow logic. CTS’s ad-delivery business runs on a set of highly specific billing rules, built up over years of contract-by-contract logic layered on top of the core platform. When Chris identified a workflow rule that needed adjusting to correctly handle a particular order type, fixing it used to mean stacking several layers of expertise: he’d describe the business requirement, a technical analyst would track down the relevant logic, and the proposed fix would get reviewed with Zuora Support before anyone touched production. It was slow, and it depended on people who weren’t always available on demand.

With Zuora AI, Chris asked where the logic lived, found the relevant rule, and worked out how to adjust it himself. He used the tool to rewrite the relevant code, tested the change in a sandbox, and only then moved the validated version into production himself. AI didn’t replace the review, Chris still knew the business rule, still defined the expected result, and still chose not to let the tool push anything to production on its own. What changed was how fast he could go from idea to a tested answer. “The speed to fix something in the time you have the idea makes all the difference,” he says. 

From assembling reports to reading them

Month-end reporting was the next place that experience got redeployed. Chris used to pull several separate reports — invoices, credit memos, payments, tax information — and stitch them together by hand to get a full picture of the period.

He asked Zuora AI to combine everything into a single query instead. The SQL it produced was more complex than anything he would have written himself, but that didn’t matter: he could compare its output against numbers he already trusted, flag where they diverged, explain what the correct result should look like, and ask the AI to adjust. It was back-and-forth validation, not blind acceptance. Once the output matched, his team had one reliable report instead of a manual reconciliation exercise.

“My team’s analyzing stuff instead of downloading and checking summaries. Those manual tasks are gone.” That shift, from assembling data to interpreting it, is central to how Chris thinks about AI’s role in finance. The goal was never to eliminate the analyst’s judgment; it was to shrink the time spent preparing information so there’s more time left to figure out what it means: “Where can I use it? How does it gain my insight into the business? That’s where my time is generally spent now.”

Guardrails, not guesswork

Chris’s own description of his approach is “more of a fail fast” philosophy — test quickly against real data, see what breaks, adjust. But fail-fast doesn’t mean careless: “I’m cautious, but a lot of times, I just want to see the results.”

He also expects the system to respect the permissions already in place. In his view, Zuora AI shouldn’t become a side door around finance controls. For example, a read-only user should stay read-only, while people with the right permissions can use AI to perform supervised actions within the bounds of their role. It’s a practical version of human-centered automation: the AI investigates, explains, queries, and executes approved work, but the finance professional stays accountable for the decision. Chris validates results, starts with tasks that have clear boundaries, and moves carefully whenever a change could touch billing outcomes.

His mental model for the tool is closer to a colleague than a machine: “You have to tell it what to do, and the more you tell it, the more accurate the result’s going to be.”

Billing as a growth lever

Chris sees a bigger opportunity past his own desk: a finance team that can help new business units stand up billing operations quickly, instead of letting billing complexity become a reason to delay or limit expansion.

He points to what changed inside his own team as the evidence. CTS’s ad-delivery business was itself onboarded with roughly 300 customers, built out one account at a time with help from Zuora’s professional services team. Today, he says, that process looks different. It’s more like: “Here is the customer list — go build me a thousand customers.”

Zuora AI gives Chris and his team another way to work through the complexity: surface the right data, investigate how the system behaves, test a fix, and move forward with more confidence.

Chris also uses other AI tools, including Cursor, when a task falls outside what Zuora AI currently handles — he treats them as complementary, not competing. The question that matters to him isn’t which tool wins; it’s whether finance is using what’s available to get better answers and cut unnecessary friction from the work.

“If anyone came to me and asked whether they should check out Zuora AI, I’d say yes, absolutely. You should leverage the tools that are available to you.”

For CTS, that means billing doesn’t have to stay a back-office process that simply gets managed. With the right controls and a finance expert still in the loop, it can become something closer to an enabler — helping the business launch, adapt, and get to revenue faster. By the end of the conversation, Chris put it more simply: “We don’t talk about billing as a negative thing anymore,” he says. And he wants to keep it that way.

Stay up-to-date on all things Zuora AI

Click here to be the first to get updates on new AI features, product releases, and upcoming events.