How DXC Technology built a multi-model AI substrate on Amazon Bedrock
Machine Learning Blog
This article describes how DXC Technology built DXC OASIS, a governed multi-model AI substrate on Amazon Bedrock for mission-critical IT operations.
- Amazon Bedrock provides unified access to multiple foundation models (Claude, Nova, Llama) with consistent governance and security controls
- Architecture separates the model substrate from the operational harness, with clear boundaries between cloud-provided and custom-built capabilities
- Resilience achieved through serverless infrastructure, guardrails, private networking via AWS PrivateLink, and human-oversight gates
- Intelligent prompt routing directs tasks to appropriate models based on cost, latency, and capability requirements
- Four authority tiers (observe, propose, act-with-approval, act-within-scope) ensure bounded autonomy for AI agents
- Across 56 measured use cases, DXC OASIS reduced resolution time by 95% on average with 92%+ diagnostic accuracy
By treating the substrate and harness as separate architectural layers with explicit boundaries, DXC created a resilient, auditable system for deploying AI agents in regulated enterprise environments.
The AWS News Feed is currently looking for gold sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.
The AWS News Feed is currently looking for silver sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.