Wood Mackenzie unites agent teams on Amazon Bedrock AgentCore

Before APEX, three Wood Mackenzie applications were each on their way to building their own agent stack: Woody, Lens AI, and the ST Trading App. Left alone, each would have stood up its own runtime, wired its own identity, bolted on its own observability, and hardcoded its own model. AgentCore resolved this by providing a managed platform rather than a library the company would need to operate itself, supporting open source frameworks such as Strands Agents and LangGraph, and models including Claude, GPT-4.1, Amazon Nova, Mistral, and Llama.

The architecture is organized in layers. The APEX frontend connects the applications to a backend where AgentCore does the work, alongside an IaC Framework for provisioning and an MCP layer for external access. A hub-and-spoke model built on AgentCore Gateway is described as the single integration pattern that does the most work in APEX: every application and agent connects through one gateway rather than maintaining its own connection to each data service, so identity, observability, evaluation, and guardrails are enforced once at the hub. APEX Studio then acts as the human-facing control plane where teams build, deploy, and monitor agents.

What the outcome hinges on is whether teams adopt that paved path rather than rebuilding pieces themselves. The platform's value rests on the same infrastructure serving both internal users like Woody and external consumers like Lens under identity-aware entitlements, and on how far multi-agent systems become the default. Whether APEX closes the gap between experimentation and production for Wood Mackenzie, and whether the shared-runtime approach spreads to other enterprise agent programs, appear to be the open questions the platform's next phase will test.