DXC Taps LangChain to Run AI Agents in Mission-Critical IT

DXC Technology is betting that the future of AI agents runs through the IT operations center. The enterprise technology firm announced on October 13 a partnership with LangChain, the company behind LangChain's open-source LangGraph agent framework and the LangSmith agent engineering platform, to bring governed agentic AI into mission-critical enterprise IT environments. The deal centers on DXC OASIS, the company's intelligent orchestration platform, which is already in production and scaling toward 125 customer environments across regulated industries and multiple geographies, according to DXC's announcement.

The numbers DXC is putting behind the partnership are striking. Across 500 measured use cases, the company says OASIS has reduced incident resolution time by 95 percent on average, with diagnostic accuracy of 92 percent or better. Those figures come from the company rather than independent testing, but they explain why a global IT services provider would standardize on LangChain's agent framework: when agents touch systems that cannot go down, the orchestration layer is the product.

LangChain's two products divide the labor

DXC was already using LangGraph inside OASIS before the partnership. The new agreement with LangChain adds a license for LangSmith, the commercial platform for agent development, evaluation, and observability. The division of labor between LangChain's two products is clean: LangGraph provides the open-source orchestration framework, including delegation between agents, state management, and human-in-the-loop workflows, while LangSmith supplies the tooling teams use to test changes and watch agents in production, according to the announcement.

That combination is aimed at a specific fear. An agent that drafts a document is easy to supervise; an agent that diagnoses a production outage, touches privileged systems, and proposes remediation steps is not. DXC says the combined platform lets it build, orchestrate, evaluate, and continuously improve agents while applying enterprise context, governance, human oversight, and continuous assurance. In practice, agents respond to operational events, investigate issues in parallel, and pull in human experts when judgment or approval is required, as described in the announcement.

OASIS itself is not new. DXC introduced it in April 2026 as an intelligent orchestration platform pairing human expertise with agentic AI, shifting managed services from reactive support toward real-time operations. As reported by Unite.AI, the platform is designed to sit across an organization's existing tools rather than replace them, with DXC's documentation naming ServiceNow and Dynatrace as examples of systems it connects to without displacing.

Why IT operations is the proving ground for agents

There is a reason the most serious agent deployments keep showing up in IT operations rather than in consumer apps. The environment is unforgiving: mistakes cascade, audit trails are mandatory, and regulated industries add compliance constraints on top. An agent framework that survives there has proven the properties every enterprise buyer asks about, which are governance, observability, and a clear escalation path to a human.

LangChain's open-source distribution gives it a structural advantage in that contest. LangGraph is already one of the most widely used agent frameworks among developers, which means DXC inherits an ecosystem of engineers who know the tooling. The commercial LangSmith license then converts that familiarity into the evaluation and monitoring layer enterprises will pay for. It is the classic open-core motion, applied to the agent stack.

The partnership also fits a broader pattern of global IT services firms becoming the distribution layer for agentic AI. These companies own the client relationships and the messy integration work that model developers cannot do at scale. GenZ NewZ recently covered a similar move when HCLTech created a dedicated unit around Anthropic's models for enterprise software work. DXC's answer runs through LangChain's open framework rather than a single model provider, which keeps its options open as the model landscape shifts.

What 95 percent faster resolution actually means

The headline metric deserves scrutiny. A 95 percent reduction in resolution time across 500 use cases is a company-reported figure, and DXC has not published the methodology behind the measurement. But the direction is consistent with what early agent deployments are showing elsewhere: the wins come from parallelizing investigation, not from replacing engineers. Agents can check logs, correlate alerts, and test hypotheses simultaneously, then hand a human a ranked diagnosis instead of a raw incident queue.

That framing matters for the labor question hanging over every enterprise AI story. DXC's pitch is that human experts focus on higher-value decisions while agents supply speed and capacity at scale. Whether that division holds as the agents improve is the open question for the entire services industry. For now, the human-in-the-loop workflow is both a safety requirement and a commercial reassurance.

The infrastructure angle is worth watching too. Agent fleets need somewhere to run, and the compute layer for the agent economy is becoming its own investment theme, as seen in recent bets on AI-agent data center hardware. DXC's partnership addresses the layer above the silicon: how agents are orchestrated, evaluated, and kept honest once they are live inside a customer's most sensitive systems.

DXC has not disclosed the financial terms of the LangChain agreement. The company's full announcement is available from DXC's newsroom, with additional reporting from Unite.AI and the PR Newswire release. If the 125-environment rollout holds, it will be one of the largest production deployments of governed AI agents to date, and a test of whether open-source agent frameworks can carry enterprise-grade accountability.