Xurrent develops software for IT service and operations teams, with a focus on IT Service Management automation. As AI takes on more repetitive tasks and increasingly gains the ability to act autonomously, the company is focused on balancing automation with human judgment, governance, auditability, and control. Pulse 2.0 interviewed Xurrent Chief Product Officer Phil Christianson to learn more.
Phil Christianson’s Background
When asked about his background, Christianson shared:
I lead product management at Xurrent, where we build software for IT service and operations teams, focused on IT Service Management automation.
Before joining Xurrent to help redefine IT service management, I led Pricing Technology at Wayfair. I’ve spent most of my career working in product roles, partnering closely with technical teams to build and deliver products.
I have always taken a deeply pragmatic approach to building software anchored in solving customer problems at every step.
Avoiding AI Complacency
When asked what is driving concern that AI may be making IT teams complacent rather than more effective, Christianson explained:
There’s a real concern that AI could make IT teams complacent, especially as it begins to take over repetitive tasks like ticket triage or alert handling. While AI can improve efficiency, the risk is that teams start relying on it to make decisions instead of supporting them. Over time, that can erode the critical thinking skills needed for more complex issues.
AI should be there to handle the repetitive work and free up time for more important issues, but when something falls outside known patterns, it’s still on the team to step in, diagnose and resolve the issue. That’s where human judgment and experience matter most.
AI As A Calculator
When asked to explain his comparison of AI for IT teams to calculators for accountants, Christianson said:
AI in IT is a lot like how calculators changed the game for accountants. It takes care of the mundane tasks, which frees up the team to focus on the important stuff. But accountants didn’t stop thinking when they started using calculators, and IT teams shouldn’t stop thinking when they use AI.
The role of the tool is to handle the mechanical work, while the team focuses on diagnosing issues and deciding on how to respond. If an IT specialist is going to use an AI agent to modify user access, for example, they should still know exactly what that AI is doing, just like you know what a calculator is doing.
Humans need to be able to validate and also be prepared to roll back changes made by AI agents.
Common AI Adoption Mistakes
When asked about common mistakes teams make when adopting AI tools, Christianson noted:
One of the biggest mistakes I see is teams rushing to adopt AI without first addressing the basics. If your knowledge base is outdated, your data isn’t clean or workflows aren’t well defined, adding AI into the mix will just amplify those problems. It’s crucial to clean up the foundational issues first.
It’s also important to measure AI’s impact both before and after implementation. Without clear metrics, like resolution time, repeated incidents or ticket volume, it’s difficult to determine whether AI is actually improving things or just complicating the process.
Additionally, it should be treated like any other IT project. Planning, communication and testing are still needed when enabling new AI-powered capabilities.
Where Humans Still Matter
When asked which tasks are best suited for AI and which should remain firmly in human hands, Christianson explained:
Like I’ve mentioned before, AI is great for handling repetitive, data-heavy tasks, like routing tickets or detecting similar incidents. These tasks are ideal for automation, freeing up time for IT teams to focus on higher-level work.
But when it comes to troubleshooting complex problems, making tough decisions, or evaluating risks, that’s where humans still need to take over.
Additionally, as Agentic AI becomes more available, organizations must look critically at where they are willing to provide agentic access. If you want to have an AI agent revoke user access in your AD, for example, the agent must store admin credentials. Where is it getting those? It’s not magic.
Beware of tools that are asking for your admins to copy/paste personal access tokens. These tools are impersonating humans in the actions they are taking and making the audit trail as well as the administration extremely difficult.
Governing Agentic AI
When invited to discuss another topic, Christianson concluded:
One topic worth mentioning is the rise of agentic AI, where systems can take action on their own. We’re seeing more AI systems that handle tasks like restarting services or scaling infrastructure autonomously.
That shifts the conversation from productivity to governance. As AI starts taking more responsibility, there needs to be clear controls in place to ensure actions are logged, auditable and reversible when needed.
AI shouldn’t operate critical systems without oversight, especially when it’s managing critical infrastructure. Every agentic action must be documented and have a rollback plan.
