Adobe’s Next Chapter: Edge Delivery Services, AI, and the Evolution of Adobe Experience Manager

Adobe is doubling down on the future of digital experience, investing in faster content delivery, streamlined authoring workflows, and AI-enabled platforms built for enterprise scale. These priorities signal a clear evolution in how Adobe expects teams to build, manage, and optimize digital experiences moving forward.

A recent two-day Adobe developer conference in San Jose brought that vision into sharp focus, offering hands-on exploration of Edge Delivery Services (EDS), deep dives into AI’s role within Adobe Experience Manager (AEM), and early insight into what’s coming next across the Adobe ecosystem.

As an Adobe architect working with large clients across multiple industries, it is helpful to know where Adobe is headed, what they are prioritizing, and how new features will benefit current and future customers, especially comparing tools like AEM to EDS.

Day 1 – Edge Delivery Services Masterclass

Evolution of Edge Delivery Architecture

The opening session introduced Adobe’s vision for EDS, a unified, low-latency delivery framework built on atomic content blocks, Git-based decentralized deployments, CDN edge computation through Adobe-managed edge workers, pre-rendered HTML stitched together at the edge, and an ultra-light JavaScript model optimized for near-zero hydration.

As an architect, watching these components work together made the entire approach feel refreshingly modern and deeply practical. It was like experiencing the simplicity of early web development again but now amplified by the global power and intelligence of today’s edge networks.

Key Architectural Features of EDS

Block-Based Rendering Model

In the block-based rendering model, content becomes blocks, each mapped to a GitHub folder containing a JS, CSS, and config file. The simplicity and transparency of this structure made me immediately want to try building a few blocks on my own.

Helix or Franklin Evolution

The 2025 unified EDS pipeline extends the Helix philosophy, letting authors use familiar tools like Google Docs or Sheets to produce production-ready HTML. This felt like a true bridge between content authors and developers.

The enhancements to the document-to-web pipeline were particularly impressive. Adobe demonstrated how semantic HTML tagging, automatic block extraction powered by machine learning, real-time preview APIs, and the new Smart Authoring Assist all work together to streamline the authoring experience. Seeing these capabilities in action made me rethink how much of our current content modeling workflows could eventually be automated. It felt like a glimpse into a future where much of the structural heavy lifting is done intelligently behind the scenes, allowing teams to focus more on creativity and less on manual content engineering.

Performance Engineering on EDS

Core Metrics Highlighted

• Sub-100 ms LCP
• <20 KB JS footprint
• Intelligent edge prefetching

Seeing these metrics achieved with minimal tuning was inspiring. It convinced me that performance is no longer an engineering task, but an architectural guarantee.

The developer-focused insights were some of the most exciting parts of the session. Experiencing a truly zero-build process, followed by instant edge updates, made the workflow feel almost unreal compared to traditional deployment pipelines. The ability to see live content mapping directly in branch previews created an incredibly fluid development loop, and the webhook-triggered auto-invalidation meant that changes propagated across the edge network in seconds without any manual intervention.

Day 2 – When AI Meets Content

Day 2 was transformative. It showed how deeply AI is now shaping Adobe’s experience stack.

Adobe presented a future where digital experiences are intelligent, adaptive ecosystems, powered by AI, semantic modeling, and edge-rendered personalization.

Key Insights and Technical Takeaways

The Rise of the Agentic Web

Adobe’s vision of the Agentic Web resonated with me most. Seeing AI agents actively assist with authoring, migration, metadata generation, and even QA felt like a glimpse into the next era of content operations. The introduction of semantic modeling, allowing AI to understand relationships between assets, components, and personalization rules, made the entire system feel far more intelligent and context aware. The real-time personalization happens directly at the edge, moving away from traditional monolithic CMS architectures toward AI-first, globally distributed experience engines. It was at this moment that I realized AEM is no longer just a CMS; it is evolving into a true intelligent orchestrator.

AEM’s Agentic Evolution

• AI-generated components
• AI-driven code refactoring
• Repository analysis
• Document-to-block transformation
• MCP and A2A for inter-agent orchestration

This made me excited to try automating parts of our migration workflows using AEM’s new agent capabilities.

Edge-Native Delivery as the New Standard

Throughout the sessions, it became increasingly clear that EDS is rapidly evolving into the default execution layer for modern digital experiences.  

The technical demonstrations were especially compelling — seeing atomic blocks rendered directly at the edge, zero-build deployments pushed from GitHub, AI-driven global cache invalidation, and millisecond-scale personalization all working seamlessly together felt like a glimpse into the future of web delivery. Even the micro-frontend compatibility showed how flexible and scalable this architecture can be for multi-brand ecosystems.  

By the end, the message was clearer to me: the edge is not an optional enhancement anymore, and it is becoming the new norm for how modern experiences should be built and delivered.

Intelligent Content Supply Chains

The discussion around intelligent content supply chains highlighted several key innovations that are transforming how content is produced and scaled. These advancements include generative content variants, automated renditions for images and videos, metadata automation driven by vision and language models, and LLM-optimized markup, designed to improve AI discoverability across channels.

Solutions like Adobe GenStudio are purpose-built to unify and accelerate the entire content supply chain—from planning and creation to delivery and analytics. By leveraging generative AI, GenStudio helps automate repetitive production tasks while surfacing real-time insights that enable teams to move faster and work smarter.

At the heart of this shift is Adobe Firefly, Adobe’s commercially safe generative AI platform. Firefly empowers creators to generate and manipulate visual, audio, and video content using natural language prompts. From text-to-image generation and advanced video editing to automated upscaling and refinements, Firefly demonstrates how generative tools can take over many of the manual, time-consuming steps in content production.

Seeing these capabilities in action made me reflect on how much more efficient multi-brand content production pipelines can become. With this level of automation reducing repetitive manual work, teams can spend less time on mechanics and more time focusing on creativity, strategy, and delivering meaningful experiences at scale.

At Bounteous, we’ve already started to instill some of this innovative efficiency into our clients’ workstreams. For example, one of our food manufacturing clients was transferring websites and needed to transfer all the recipes hosted on the old website to the new website. By training AI on the AEM content fragment model, we were able to transfer and upload all the recipes automatically, with no custom code required. This took what would have required hundreds of hours of manual labor to just 1 to 2 working days.

Developer Experience: Faster, Leaner, & More Autonomous with AI

AI is radically transforming how developers approach their work, and seeing these capabilities in action was genuinely energizing. From AI-assisted coding and debugging to automated runtime optimization and edge-first performance debugging, the entire development cycle felt noticeably lighter and more intuitive.  

The introduction of micro-frontend templates added another layer of flexibility, making it easier to structure and scale multi-brand architectures. As someone who has spent years wrestling with heavy, legacy build systems, experiencing this new level of automation and simplicity felt nothing short of liberating.

Search, Retrieval, and AI-Optimized Experiences

The combination of AI agents and search engines like Algolia offered:

• Semantic and vector retrieval
• Hybrid search
• Unified content graph indexing
• LLM-optimized site markup

I’m particularly excited to test how EDS output performs in LLM-focused search engines.

AEM vs. EDS, Reframed

Before attending this event, I used to think of AEM and EDS as two separate strategies: AEM for traditional enterprise CMS needs and EDS for high-performance, lightweight sites.

Now, I see AEM as the intelligent orchestration and content brain, EDS as the high-performance edge-native execution layer, AI agents as the automation engine connecting the two, and semantic blocks as the unified structure that ties authoring and delivery together.

I left the event feeling energized and inspired, eager to experiment with everything from AI-driven migrations to edge-native personalization and block-based architectures. For me, it was clear that the future of digital experience engineering is agentic, edge-native, and powered by automation.