About
Sophos is a global cybersecurity leader, protecting organizations from advanced threats with AI-native endpoint, network, email, and cloud security and managed detection and response. Headquartered in the UK, Sophos expanded its portfolio with the acquisition of Secureworks.
01
The Challenge
Sophos ran a highly customized billing and revenue environment. Every new data correction meant building and testing another workflow or preparing a CSV file by hand, and the finance team relied on technical colleagues to write queries for the data they needed. As a security company, Sophos also set a very high bar for moving financial data into third-party AI tools.
02
The Solution
Sophos enabled Zuora AI across its Billing and Revenue tenant, with read-and-write access in production governed by existing user permissions. Teams now update records, query data, build dashboards, and document workflows in plain language. No more waiting around for scripts, or sharing sensitive financial data with third parties.
03
The Benefits
“We tried the prompt in QA, then used exactly the same prompt in production, and it all worked. It saved all the development effort and all the testing effort. I’d put it at around 85 percent time saved.”
Fixing billing data at scale used to follow the same pattern at Sophos: prepare a CSV file with exactly the right record IDs, field names, and values, then run it through one of several custom workflows. A new scenario meant a new workflow, with development, testing, and approvals.
With Zuora AI, the team simply describes what it needs. It asks Zuora AI to find the records that need attention, reviews the list, and tells it what to change. The finance team gets answers directly, instead of waiting on developers.
The impact was clearest during one month-end close, when an issue in an upstream system left a large volume of records needing correction. Rather than build a new workflow against a fixed deadline, Deepak’s team tested a prompt in QA, confirmed the results with business users, and ran the same prompt in production.
Many of Sophos’s workflows were built by an implementation partner and came without documentation, a real risk for business continuity and for onboarding new team members. Now Deepak gives Zuora AI a workflow’s name or definition and receives complete, readable documentation in return.
The team also builds dashboards directly from billing data, including one that tracks e-invoicing volume, accuracy, and turnaround time across three countries. Deepak can also see which prompts colleagues rely on, and save the best ones as runbooks the whole team can reuse.
“Zuora is the source. Why should I extract the data and send it to another application to do the same prompting, when Zuora AI can build the whole dashboard?” — Deepak Singh, Sophos
For a company whose business is protecting data, where AI runs matters. Connecting financial data to an external AI tool at Sophos used to require compliance review, connector approvals, and additional licensing, as well as the risk of data leaving a controlled environment. Zuora AI works where the data already lives.
“If I’m extracting data outside, there’s a risk of data loss. If I’m only working in Zuora, the data stays within those fences.” — Deepak Singh, Sophos
Sophos runs Zuora AI with read-and-write access in production, with permissions that follow each user’s existing role. The team keeps a person in the loop for every change: prompts are tested, results are reviewed, and nothing is trusted blindly. “It was very easy to configure from day one,” Deepak says.
Sophos is building a knowledge base of its workflow documentation and customizations so Zuora AI can answer questions specific to how Sophos runs its business, and it is looking ahead to scheduling recurring analyses and dashboards directly in Zuora.
“Every time I think I should have a feature, the Zuora AI team comes back and says, ‘Here it is, this is coming up next.’” — Deepak Singh, Sophos
His advice to peers is simple: “It’s as good as any other AI tool. Just go and try it.”