News
DXC's AI Agents Target the Wait Behind IT Incident Resolution
When an enterprise system develops a problem, the time spent fixing it can include waiting for someone to investigate, locating the right specialist, and securing approval to act. AI agents promise to shorten that process. For IT leaders, the question is how much work they can safely delegate.
DXC Technology’s October 13 announcement offers a company-reported example. The technology services provider says its OASIS platform reduced IT issue resolution times by an average of 95%, with diagnostic accuracy of at least 92% when compared with senior engineers’ judgments.
The potential significance extends beyond faster troubleshooting. If agents can investigate problems while engineers handle other work, organizations could reduce delays without requiring a specialist to initiate every investigation. But permission to investigate a system is different from permission to change it.
Changing How Work Reaches Engineers
In an accompanying
technical white paper, DXC and LangChain describe an operating model in which events can trigger agents, which then delegate investigations to specialized agents working in parallel.
That approach targets a familiar constraint: technical expertise is distributed across teams, while incidents can involve several systems. The proposed benefit is less time spent waiting between those areas of expertise.
For an IT manager evaluating the approach, an important distinction is whether reported savings come from faster technical analysis, shorter queues, or both. That affects where a similar deployment might help their organization.
Speed Requires Boundaries
DXC says OASIS assesses actions according to factors including potential impact and reversibility. Sensitive actions can pause for human approval, while decisions and tool interactions are recorded.
Those controls matter because an agent investigating a database slowdown and an agent changing a production database carry different operational risks. Buyers need to understand where autonomous investigation ends and approval becomes mandatory.
The reported diagnostic accuracy also measures something narrower than overall operational safety. Agreement with an engineer’s diagnosis does not, by itself, establish that a proposed remedy was authorized, successfully executed, or free of unintended effects.
What the Results Need to Establish
The headline reduction is a reason to examine the approach, rather than a forecast of what another organization will achieve. Useful evaluation would establish when the resolution clock starts and stops, how unsuccessful attempts are counted, and whether human review is included.
For enterprise readers, the consequential question is whether agents can shorten service disruptions and reduce engineering workload while leaving the organization able to verify and control their actions.