AI agents that operate Adobe Experience Manager — not as demos, but as production systems running across 100+ markets and serving millions of users.

I make enterprise platforms operable by AI agents. That means building the protocol layer (MCP servers), the orchestration layer (multi-agent workflows), and the autonomous execution layer (24/7 pipeline agents) — all designed for the constraints enterprises actually have: governance, multi-environment deployments, team-scale coordination.

My systems handle the full development lifecycle autonomously: requirements analysis → implementation planning → code generation → multi-phase verification → PR creation. One command. No re-explaining the project each session.

This isn't research or prototyping:

• 100+ market AEM platform — agentic workflows operating across a multi-market content platform
• Millions of users — systems deployed in high-traffic enterprise e-commerce environments
• Autonomous pipeline agents — AI agents running 24/7 as Azure DevOps pipelines, triggered by webhooks, producing verified PRs without human intervention
• Multi-agent orchestration — Opus for deep review, Sonnet for execution, Haiku for lookups — tiered by task complexity

Years of open source and internal enterprise tooling — from UI components and developer utilities to deployment automation and platform SDKs. Recently shifted focus to agentic workflows:

The protocol layer. MCP server that gives AI agents direct access to AEM — JCR content, components, dialogs, page operations. The interface between LLMs and enterprise CMS.

The orchestration layer. 70+ skills running identically across Claude Code, GitHub Copilot CLI, and VS Code Chat. Full development lifecycle from ticket to PR — config-driven, never hardcoded. Includes autonomous agents for DoR/DoD validation, code review, bug fixing, and QA.

GitHub Action for SSH deployments. 1k+ stars. Used across thousands of CI/CD pipelines.

Contributor to Adobe's official AEM Headless SDK.

...and many many more.

• Agentic workflows — LLM + tools + structured execution, not chatbots
• MCP (Model Context Protocol) — building the interface layer between AI agents and enterprise systems
• Enterprise AI constraints — governance, verification gates, multi-environment deployments, team coordination
• CI/CD and developer infrastructure — deployment automation, pipeline design, developer tooling
• AEM platform engineering — deep specialization in making Adobe Experience Manager AI-operable

Expanding the surface area of what AI agents can operate in enterprise environments. More MCP servers, deeper orchestration, more autonomous pipelines. The goal: enterprise development workflows where AI agents handle execution end-to-end, with humans setting direction and reviewing output.

Berlin · @draganfill · LinkedIn