I make enterprise codebases agent-friendly — and rewire how teams ship software around coding agents.
AI agents that operate Adobe Experience Manager — not as demos, but as production systems running across 100+ markets and serving millions of users.
1,300+ GitHub stars · 3 Adobe-distributed SDKs · 57 MCP tools · 78 AI skills
I embed with engineering teams and make their codebases agent-friendly — building the rules, skills, and MCP servers that let AI coding agents actually work in large, complex enterprise codebases. Then I redesign the SDLC around them: finding where delivery breaks when you add agents (QA bottlenecks, poorly scoped requirements, review debt) and rewiring it — and bringing the whole team up, not just the engineers already excited.
Under that sits the tooling I've built for the constraints enterprises actually have — governance, multi-environment deployments, team-scale coordination:
- the protocol layer (MCP servers),
- the orchestration layer (multi-agent workflows),
- the autonomous execution layer (24/7 pipeline agents).
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:
- Daily in production on a Fortune 500 account — a 13-agent requirements-to-PR platform (KAI) running every day, escalated to COO
- 100+ market AEM platform — agentic workflows operating across a multi-market content platform
- Millions of users — systems deployed in high-traffic enterprise 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. 57 MCP tools that give AI agents direct access to AEM — JCR content, components, dialogs, page operations. The interface between LLMs and enterprise CMS. One of the first full MCP servers for AEM.
The orchestration layer. 78 skills, 13 agents, 4 plugins 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. Shipped ~6 months before Salesforce productized the "Agentic Engineering" category.
AI parliament simulation. 6 autonomous agents model German political parties using Claude + Grok. Multi-provider orchestration, circuit breakers, and coalition negotiation running a full 4-year parliamentary term for ~$44. This is the project that got me found — a recruiter discovered it via GitHub search and reached out directly (June 2026).
GitHub Action for SSH deployments. 1,300+ stars, 162 forks, 57 releases. Used across thousands of CI/CD pipelines.
Zero-config Node.js mock API server with web GUI. File-based endpoint management, dynamic path parameters, response variations. Built to unblock frontend teams waiting on backend APIs.
All-in-one frontend toolchain for AEM projects. Standardized build processes adopted across Netcentric, setting company-wide frontend standards.
Lead contributor to Adobe's official AEM Headless SDK for JavaScript/Node.js.
- Making codebases agent-friendly — rules files, custom skills, MCP servers; context engineering for large, complex repos
- SDLC redesign & enablement — rewiring delivery around coding agents, evals/LLMOps, bringing whole teams up
- 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
- AEM platform engineering — deep specialization in making Adobe Experience Manager AI-operable
- adaptTo() 2023 — "AEM Headless: A Glimpse of Developer Tools" (Europe's leading AEM conference)
- Frontend Coffee Break — Podcast Episode #25: "Open Sourcing Our FE Build"
- Netcentric Blog — "AI-Driven Code Generation: AEM Component Development from Figma" (2025)
- Multiple internal/client talks — AI-enabled development, AEM headless, frontend architecture (audiences up to 200+)
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 · frontenddot.com