AWS Launches Agent Registry in Preview to Govern AI Agent Sprawl Across Enterprises. AWS launches an Agent Registry preview to centrally discover, govern, and reuse enterprise AI agents with native MCP and A2A protocol support. Outcome engineers get a control plane for agent lifecycle—discovery, reuse, and policy enforcement become engineering primitives (Principles 09 & 10).

Agents That Remember: Introducing Agent Memory. Cloudflare launches Agent Memory beta: a managed retrieval-based persistent memory that lets agents recall context without filling model context windows. Outcome engineers can externalize long-term state to a reproducible service, simplifying context engineering, testing, and predictable agent behavior (Principles 06 & 11).

Anthropic Introduces Agent-Based Code Review for Claude Code. Anthropic ships agent-based Code Review for Claude Code that runs parallel AI reviewers to find, verify, and prioritize pull-request issues in about 20 minutes. This turns agentic QA into an auditable stage of your delivery pipeline and models a pattern where agents produce artifacts and other agents verify them (Principles 03, 09 & 15).

OpenAI ratchets up Codex’s agentic capabilities to rival Claude Code. OpenAI upgrades Codex with agentic capabilities to automate complex coding tasks and compete with Anthropic’s Claude Code. For outcome engineers this accelerates the shift from human-in-the-loop tooling to agent-first workflows, forcing new orchestration, verification, and safety patterns (Principles 09 & 03).

Introducing Flagship: Feature Flags Built for the Age of AI. Cloudflare releases Flagship, an OpenFeature-based, edge-native feature flag system designed for agent-driven deploys so teams can test, roll out, and rollback code at the edge. Feature flags become the safety gate and experiment harness for iterative agent deployment and outcome validation in production (Principles 01 & 15).