Building the agentic cloud: everything we launched during Agents Week 2026. Cloudflare launches a suite of agent-first primitives—sandboxes, Git-compatible Artifacts, identity-aware egress, Durable Object Facets, and Workflows v2—to run agents at edge scale. Outcome engineers get a concrete platform blueprint for safe execution, artifact provenance, and identity-aware egress that map directly to runtime and orchestration patterns (Principles 06, 07, 09).

The AI engineering stack we built internally — on the platform we ship. Cloudflare describes building its internal AI stack on the same platform it ships, integrating an AI Gateway, Workers AI, Agents SDK, Sandbox SDK, Workflows, and Backstage to route billions of tokens. This is a production case study showing how to turn agent primitives into developer tooling, observability, and operational controls you can emulate in your org (Principles 06, 07, 14).

Orchestrating AI Code Review at scale. Cloudflare implements a CI-native orchestration that runs specialized AI reviewers, deduplicates findings, and blocks unsafe merges across thousands of pull requests. If you’re embedding agents into developer workflows, this shows the pattern for reviewer specialization, deduplication, and CI gating to keep agent output actionable and auditable (Principles 09, 06, 15).

Google ADK for Java 1.0 Introduces New App and Plugin Architecture, External Tools Support, and More. Google ships an ADK with app/plugin architecture, external tool integrations, and human-in-the-loop context engineering targeted at agent development. This materially lowers the friction of building tool-enabled Java agents and formalizes context and plugin interfaces you’ll want to mirror for robust, auditable agent behaviors (Principles 03, 06, 15).

Kimi K2.6: Advancing Open-Source Coding. Kimi open-sources K2.6, a coding-focused model that excels at long-horizon, tool-enabled agent workflows and multi-agent orchestration. New model capability changes tradeoffs for orchestration and planner design—expect fewer brittle hand-offs and more reliable long-run plans when you rearchitect agents around K2.6-style strengths (Principles 03, 09).