AEM Champions Office Hours: 2025 Wrap-Up and the Path to Agentic AEM (December 2025)

The last AEM Champions Office Hours of 2025 brought together a panel of AEM Champions, including RENY director Rami El Gamal and Brett Birschbach, with host Greg Dimeris. They looked back at the year's biggest AEM moments and talked through what agentic AEM means for authors, developers and product owners in 2026.

What the session covers

• Adobe's agent lineup: Agent Orchestrator, LLM Optimizer, Sites Optimizer, Brand Concierge and AEM answers, plus MCP services and the expanded AEM OpenAPI and events APIs.
• What stood out in 2025: many AI "seeds" were planted, but most are still evolving, and clients asked how to get to AEM as a Cloud Service so they could use them.
• Practical adoption: picking the agents that fit a project, and filling the gaps with custom AI tooling.
• AI visibility: how LLM Optimizer shows what AI crawlers see, and why AI answers change how sites earn traffic.
• Authoring and personalization: Document Authoring versus the Universal Editor, and whether personalization still matters.
• Content quality: how search engines and LLMs might treat AI-generated content, and why human validation stays essential.

Key takeaways

• Defining the problem is the hard part. Everyone asks whether AI can do it, but the hardest part of any implementation is aligning business units on what "it" is. The technology is often ahead of the business requirements.
• Cloud Service is the entry ticket. Many agentic features depend on AEM as a Cloud Service and Edge Delivery Services, so teams still on-premise or on Managed Services risk arriving late.
• You don't need every agent. Agents tend to have narrow use cases. Choose the ones that improve productivity on your project, and check their dependencies (some asset agents work best with Dynamic Media, for example).
• Small wins count. Removing repetitive work from authors, such as alt text, translations or related-product suggestions, delivers value now, even if it is less exciting than replacing a whole process.
• Combine agents with your own tooling. One panelist's team paired Adobe's site migration agent with custom scripts for complex implementations the agent could not handle alone.
• AI visibility needs a business view. LLM Optimizer can show the difference between what people and AI crawlers see, but knowing which questions your audience asks AI tools is still the harder part. Server-rendered HTML helps LLMs read your content.
• AI answers can reduce visits. If your business relies on visitors reaching a form, being summarized in an AI answer is not automatically a win. Measure what matters to your funnel.
• Validation is not optional. As AI makes it easy to generate thousands of pages, the temptation to publish without review grows. Content management and governance matter more, not less.

RENY's perspective

We see 2025 as a year of seeds that have not fully grown yet: there is plenty of talk about MCP servers and agents, but businesses first need to define what they want AI to do, because aligning business units on requirements is harder than the implementation itself. Other panelists agreed, noting that much of the agent documentation still describes work in progress, and that the clients moving fastest were those already planning their move to AEM as a Cloud Service.

How RENY helps

We help teams get ready for agentic AEM: moving to AEM as a Cloud Service and Edge Delivery Services, choosing the AI features that fit, and putting content governance in place before scaling AI-generated content. Talk to an AEM architect.