DXC Technology launches multi-model AI system DXC OASIS on Amazon Bedrock for enterprise IT operations
DXC Technology launched (GA March 2026) the DXC OASIS system on Amazon Bedrock, which orchestrates AI agents across the Claude, Nova, and Llama models with governance controls. According to DXC, it reduced IT incident resolution time by an average of 95% across 56 use cases.
DXC Technology, a provider of managed IT services for banks, hospitals, and government institutions, reached general availability of the DXC OASIS system in March 2026 — a layer for orchestration, governance, and auditing of AI agents in enterprise IT operations. The system is built on Amazon Bedrock as a "model substrate," which gives agents unified access to multiple models at once: Claude from Anthropic, Amazon Nova, Meta Llama, and other open-weight models.
According to the article, Bedrock Guardrails provide unified content filtering, redaction of personal data, and enforcement of prohibited topics across all models used. Inference traffic is routed exclusively within the customer's virtual network (VPC) via AWS PrivateLink, and according to DXC, prompts and outputs are not shared with model providers or used for their training. Thanks to the connection via Bedrock, an outage or withdrawal of one model becomes, according to DXC, a matter of redirecting traffic rather than rebuilding the entire system.
According to DXC, since reaching general availability in March 2026 the system has reduced resolution time by an average of 95% across 56 tracked use cases. The company plans to expand to approximately 130 customer environments running on Amazon EKS with separate namespaces for individual customers. DXC Technology operates IT services with more than 115 000 employees in 70 countries.
Further technical details of the architecture and a list of specific resilience controls are not included in the available part of the source article — see the source article for details.
Why it matters
For companies in regulated industries, this illustrates a concrete architectural pattern for deploying AI agents with greater autonomy without dependency on a single model provider and with built-in controls for governance, data isolation, and auditability — addressing risks that have in practice emerged with autonomous agents acting beyond their intended scope. However, the stated 95% reduction in resolution time is a vendor's claim about its own product based on a limited sample of 56 use cases, not an independently verified result.
Relevant practical impact
What this means
For a business
Companies in regulated industries (banks, hospitals, government) that use DXC's managed IT services gain an AI agent system with built-in governance (content filtering, PII redaction, data isolation within the customer's VPC) and without dependency on a single model provider, which according to DXC reduces outage risk and speeds up incident resolution.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 0 independent. We count feeds from the same owner only once.