DXC and LangChain advance a proxy-based AI framework for critical IT operations and maintenance
This collaboration will combine the decades of experience of DXC in operating critical enterprise systems with the technology of LangChain used for security management of the AI proxy framework.
DXC OASIS, built on the open-source LangGraph runtime based on LangChain, is now being extended to 125 customer environments, covering regulated industries and multiple regions.
In 500 measurement use cases, OASIS can reduce resolution time by an average of 95%, with a diagnostic accuracy rate of 92% or higher.
Ashburn, Virginia, October 13th ( PRNewswire ) – DXC Technology ( NYSE : DXC ) today announced a partnership with LangChain. LangChain is the company behind the open-source LangGraph proxy framework and LangSmith proxy engineering platform. Both parties will advance governed proxy-based AI for application in critical enterprise IT environments.
This collaboration is built on the foundation of LangGraph's usage within DXC OASIS by DXC. OASIS is an intelligent orchestration platform and operational model developed by DXC for critical IT operations and maintenance. DXC also continues to be licensed for use with LangSmith to support the development, evaluation, and observability of OASIS agents. OASIS has now been put into production and is being expanded to 125 customer environments, covering regulated industries and multiple regions. Both parties state that these capabilities help DXC to build, orchestrate, evaluate, and continuously improve AI agents, while also applying the necessary corporate context, governance, human supervision, and continuous assurance in key technical environments.
Introducing Proxy AI into Critical Task Operations and Maintenance
Through this collaboration, DXC is utilizing LangGraph and LangSmith to advance the construction, orchestration, evaluation, and continuous improvement of OASIS proxies. LangGraph provides an open-source framework for proxy orchestration, including capabilities such as delegation, state management, and human-in-the-loop workflows; LangSmith offers tools for proxy development, evaluation, and observability.
OASIS integrates these capabilities into the enterprise IT operations, enabling agents to respond to operational events, investigate issues in parallel, leverage specialized capabilities across technical domains, and involve personnel in the process when judgment or approval is required. DXC indicates that this allows human experts to focus on more value-added decisions, while agents provide the capacity and speed necessary to handle large-scale operational issues. In specific production use cases, file system events that previously took 53 to 90 minutes to resolve can now be solved within 3 to 6 minutes; database performance issues have been reduced from 2 to 3 hours to 5 to 12 minutes; and complex issue management, which used to take 2 to 6 days, has now been shortened to 8 to 10 minutes.
DXC Technology, President of Global Infrastructure Services, stated: "Proxy-based AI has the potential to fundamentally change the way the most critical technologies in enterprise operations are used, but models alone are not enough. Based on over 500 measurement use cases, OASIS has reduced average resolution times by 95%, and when compared with judgments verified by senior engineers, the diagnostic accuracy rate reaches 92% or higher. Through collaboration with LangChain, we are combining advanced proxy technology, operational experience, and governance to ensure that AI proxies can be safely implemented on an enterprise scale."
Expand proxy AI under governance and human supervision.
One of the key focuses of this collaboration is to help enterprises expand the use of AI agents while maintaining proper governance and human supervision. With the help of LangChain technology, OASIS can now allow agents to operate within established permissions while still maintaining the control required in critical task environments. The system evaluates actions based on factors such as reversibility, impact, verifiability, and the agents' past performance. Sensitive operations can be paused before execution to await human approval; at the same time, agent decisions, tool calls, and results are all tracked to provide a clear understanding of how the agents function.
OASIS also allows agents to make inferences based on the specific corporate environment they are in. Agents can invoke system-specific and customer-specific information, DXC operational experience, as well as supplier documentation, in order to make judgments according to the context and requirements of each customer's environment, rather than relying solely on general model knowledge. DXC indicates that by combining corporate context with clear governance policies and human supervision, OASIS organizations are able to expand the role of AI agents in a production environment while maintaining accountability and control.
LangChain Chief Operating Officer Julia Schottenstein stated: "Proxies are reshaping the application layer of every enterprise, while leading companies are taking control of their own proxy technology stacks. This means adjusting frameworks to suit their own business needs, enabling proxies to be rooted in proprietary contexts, and establishing governance and continuous improvement cycles to ensure reliable operation in production environments. We are delighted to bring this model to key enterprise operations through our collaboration with DXC."
Open path for enterprise-oriented proxy AI
This collaboration also supports an open proxy-based AI approach, aimed at enabling enterprises to maintain flexibility and control throughout the technological evolution of AI. OASIS is not bound to a single model and supports interoperability with third-party systems as well as those owned by customers, allowing organizations to choose different models based on requirements such as quality, cost, latency, and privacy. OASIS is also designed to continuously improve during operation, leveraging insights gained from proxy execution to enhance performance, and to incorporate experiences from individual scenarios into a broader system over time.
Deployment can cover public clouds, private clouds, dedicated environments, and sovereign environments without the need to redesign the underlying architecture. Customers retain control over their data, workflows, and operational capabilities built on top of the platform, while maintaining flexibility between models, infrastructure, and deployment environments.
DXC and LangChain detail their approach in a new technical white paper titled "DXC OASIS : The Harness for Mission - Critical IT Operations". This white paper explores how organizations can deploy AI agents into actual operations while maintaining governance, accountability, and human supervision in critical task environments.
About DXC Technology
DXC Technology ( NYSE : DXC ) is a leading enterprise technology and innovation partner that provides software, services, and solutions to global enterprises and public sector organizations, helping them drive results in an era of rapid change with AI. With its deep experience in managed infrastructure services, application modernization, and industry-specific software solutions, DXC modernizes, secures, and operates some of the world's most complex technical systems. For more information, please visit dxc.com.
