Article | September 17, 2026
By Lisa Beaudoin, Chief Product Officer, DXC Technology
Most technology leaders have already accepted a hard truth. Agentic AI is arriving in IT operations, whether or not the operating model is ready for it.
The drive toward faster resolution and lower toil is obvious, and so is the hesitation. Gartner expects more than 40% of agentic AI projects to be canceled by the end of 2027, often because of unclear business value or inadequate risk controls. Only 17% of organizations have deployed AI agents so far, yet more than 60% plan to within two years.
The harder question is narrower. Leaders need agents that can act inside mission-critical environments (the systems that run banks, airlines, hospitals and government services) without losing sight of what those agents are doing or why.
That’s the design problem DXC set out to solve with DXC OASIS, an agentic IT operations platform that pairs human decision-making with AI execution across the entire IT estate.
Agentic operations are often confused with automation, but there’s a clear distinction:
Automation runs a script when a condition is met.
Agentic operations combine decisions and actions produced by people and AI working together, so the system can interpret a situation, weigh options and act, then hand the judgment call back to a person when it counts.
In practice, a master agent coordinates specialized sub-agents across the estate. Each handles part of the picture (monitoring, correlation, remediation) while the master agent keeps the work aligned to the outcome the business actually cares about.
None of this runs as a black box. Every action is traceable and every insight explainable, which is the difference between a tool leaders tolerate and one they trust in a crisis.
The most important design choice in DXC OASIS could be easy to overlook. The platform is built to amplify people rather than replace them. Agents absorb the volume of routine, repetitive work so engineers can move from reacting to alerts to leading their operations and getting ahead of problems. Operators engage the platform conversationally, in plain language, rather than navigating dashboards.
That framing came from research, not assumption. Before making design decisions, DXC studied how operators and customers actually work and where friction shows up, then validated those findings at scale. The team uncovered six recurring disconnects, from limited business insight to tools that trap staff in reactive mode, and designed agents around the people doing the work rather than around isolated tasks.
Human assessment stays central to every decision loop, which is what a VP of infrastructure or a head of IT operations needs to hear before handing any authority to software.
DXC OASIS sits atop the tools an enterprise already runs, connecting environments like ServiceNow, Dynatrace and everything else you run into a single, real-time view without ripping anything out. That sole control plane is where AI agent orchestration happens. Signals from across the estate are pulled together, patterns are read and risks are forecast before they reach the business, so teams can act on what’s coming instead of only reacting to what’s broken.
The outcomes are reflected in the numbers. Routine tasks that used to take about 1.5 hours are now completed in 3 to 6 minutes. Complex problem management, once handled by pulling experts into a war room, is increasingly coordinated through AI agent orchestration, with 99% reduction in time and around 92% first-time accuracy.
For global operations that run around the clock, that shift from manual coordination toward continuous, governed execution is the point.
For a CISO, speed means nothing without control, so governance is built into the model rather than bolted on as an afterthought. AI agent governance defines how agents are certified before they’re allowed to operate, what they’re permitted to do and where a human must sign off. AI agent observability then makes an agent’s behavior visible in real time, showing not only what an agent did but also why it did it.
This is also the honest answer to the worry about hallucinations. High first-time accuracy combined with full traceability means an agent’s reasoning can be inspected and its actions audited rather than taken on faith. That matters well beyond IT, given that regulators across the United States, Europe and Asia Pacific are moving quickly on AI accountability, and explainability is becoming a compliance requirement rather than a nice-to-have.
Plenty of vendors can build agents. Far fewer run the mission-critical estates that those agents are meant to manage. DXC has operated the world’s most complex systems for decades, informed by insights from thousands of customers and millions of devices. In addition, we proved DXC OASIS within its own operations first, using a customer zero approach before offering it to clients.
The platform’s AI capabilities are deepened by a multi-year global alliance with Anthropic, whose Claude models serve as the default foundation for DXC OASIS agentic workflows and have helped accelerate the platform’s own software delivery roughly tenfold. DXC is also training forward-deployed engineers, certified in 90 days, to bring these capabilities directly into customer environments under the same security and compliance standards those customers face.
Agentic projects that fail tend to lack clear value and real controls. The ones that compound put governance and observability in place from day one and prove themselves on work that really matters.
Leaders should begin with a bounded, mission-critical domain in which visibility and control are already hard to maintain, insist on traceable decisions and trackable outcomes and choose a partner that operates the estate rather than one that only supplies the software.
Do that, and agentic AI stops being a risk to manage and becomes a way to run operations with more speed and advanced control.
Lisa Beaudoin is Chief Product Officer at DXC Technology, spearheading the product strategy and engineering for innovative AI-driven enterprise platforms. She operates at the intersection of technology, user experience, and data to develop breakthrough products. Connect with Lisa on LinkedIn.
