DXC Taps LangChain Tools to Run AI Agents in Mission-Critical IT
DXC Technology has announced a partnership with LangChain to advance governed agentic AI in mission-critical enterprise IT. DXC says its OASIS platform, built on LangChain's open-source LangGraph runtime, is already in production and scaling toward 125 customer environments.
- DXC Technology has partnered with LangChain, and OASIS is built on the open-source LangGraph runtime.
- DXC is licensing LangSmith for development, evaluation and observability of OASIS agents.
- OASIS is in production and scaling toward 125 customer environments, according to DXC.
- DXC claims a 95% average cut in resolution time across 500 measured use cases, with diagnostic accuracy of 92% or better.
- Sensitive agent actions can be paused for human approval, and decisions and tool calls are traced.
What the partnership covers
LangChain is the company behind the open-source LangGraph agent framework and the LangSmith agent engineering platform. According to DXC, the partnership builds on its existing use of LangGraph within DXC OASIS, its intelligent orchestration platform and operating model for mission-critical IT operations. DXC is also licensing LangSmith to support the development, evaluation and observability of OASIS agents.
LangGraph supplies the agent orchestration layer, including delegation, state management and human-in-the-loop workflows. LangSmith provides tooling for development, evaluation and observability. DXC says OASIS integrates these capabilities so agents can respond to operational events, investigate issues in parallel, draw on specialized capabilities across technology domains, and bring people in when judgment or approval is required.
OASIS is already in production and scaling toward 125 customer environments across regulated industries and multiple geographies, according to DXC.
Performance claims from DXC
DXC reports that across more than 500 measured use cases, OASIS has reduced resolution time by 95% on average, with diagnostic accuracy of 92% or better when validated against senior engineer judgment. These figures come from the company itself.
The company also cited specific production examples:
- Filesystem incidents that previously took 53-90 minutes can be resolved in 3-6 minutes.
- Database performance incidents have moved from 2-3 hours to 5-12 minutes.
- Complex problem management that previously required 2-6 days has been reduced to 8-10 minutes.
"Agentic AI has the potential to fundamentally change how enterprises operate their most critical technology, but models alone are not enough," said Dan Gray, President, Global Infrastructure Services, DXC Technology.
Governance and human oversight
A central focus of the partnership is letting enterprises expand the use of AI agents while keeping governance and human oversight in place. Using LangChain technology, OASIS now lets agents operate within defined authority. Actions are assessed on factors including reversibility, impact, verifiability and an agent's track record. Sensitive actions can be paused for human approval before execution, and agent decisions, tool calls and outcomes are traced to show how the agents operate.
OASIS also grounds agents in the specific environments they work in. Agents can draw on system-specific and customer-specific information, DXC operational expertise and vendor documentation. DXC says this helps them reason using the context of each customer rather than relying solely on general model knowledge.
Julia Schottenstein, Chief Operating Officer at LangChain, said enterprises pulling ahead are "taking ownership of their agent stack", which includes adapting the harness to their business and building governance and continuous improvement loops.
An open, model-agnostic approach
DXC describes OASIS as model agnostic, with interoperability with third-party and customer-owned systems. Organizations can choose different models for different workloads based on quality, cost, latency and privacy. The platform is designed to improve as it operates, using insights from agent executions and incorporating lessons from individual scenarios into the wider system over time.
Deployments can span public cloud, private cloud, dedicated and sovereign environments without redesigning the underlying architecture. According to DXC, customers retain control of their data, workflows and the operational capabilities they build on the platform.
What it means for enterprise IT
The announcement shows a services provider pairing an open-source agent framework with an evaluation and observability tool to run agents in regulated, high-stakes settings. The emphasis on approval gates, tracing and customer-specific context reflects the controls that operators of critical systems are likely to require. The performance numbers are DXC's own and the release does not describe how they were independently verified.
DXC and LangChain have also published a technical white paper, DXC OASIS: The Harness for Mission-Critical IT Operations, which explains how organizations can operationalize AI agents while maintaining governance, accountability and human oversight.
