This is a part two in a three-part series covering the innovations and new interfaces unveiled at this year’s Adobe Summit 2026. Part one focuses on how Adobe is reshaping B2B marketing through AI decisioning and agents.
At Adobe Summit 2026, one phrase kept surfacing across keynotes and breakout sessions: the "agentic enterprise." It signals a fundamental shift in how AI is expected to operate inside a business, moving from isolated tools that generate outputs to interconnected agents that take action.
In part one of this series, I covered how Adobe is reshaping B2B marketing with AI agents and decisioning. Here, I’ll go deeper into the architectural framework Adobe introduced to help organizations make that shift: the agentic capability stack.
In systems thinking, a "leverage point" is a place within a complex system where a small shift produces a massive, systemic change. When companies struggle with AI adoption, it is usually because they are intervening at the wrong leverage point — focusing on surface-level outputs rather than deep, structural inputs.
To guide businesses through this transition, Adobe introduced Adobe CX Enterprise, an end-to-end agentic AI system designed to manage the entire customer lifecycle. It’s anchored by the foundational framework of the agentic capability stack, which helps organizations architect the shift to an agentic enterprise safely and effectively.
Based on your organization’s own AI maturity, here is how marketing teams can leverage this framework to build an AI agent ecosystem.
Adobe's research shows that 75% of organizations cite data integration and quality as their top AI implementation challenge. That makes AI-ready data the foundational layer of the stack and a critical starting point for any marketing team. For an AI agent to take meaningful action, it needs a trusted, structured, and machine-readable data foundation.
In practice: This is where Adobe Real-Time CDP serves as the contextual brain. To ensure agents have the complete picture, Real-Time CDP now supports unstructured data, turning raw intent from call center logs, chat transcripts, and video interactions into machine-readable vector embeddings.
It's worth noting that vector embeddings and unstructured data support are relatively new capabilities. Most organizations will need to assess whether their current data pipelines are structured in a way that can actually feed them.
Once the data is structured, the next layer requires journey rails. These are clearly defined, API-driven discovery and transaction rails that enable agents to move customers smoothly from start to finish. Agents need to understand the pathways a customer can take before they can optimize them.
In practice: Adobe Customer Journey Analytics (CJA) can provide these rails by mapping out complex buyer paths and measuring what truly drives demand across every channel. By clearly defining these pathways, agents can accurately forecast outcomes and allocate resources effectively.
Before granting autonomy to AI, organizations must ensure their Integration and governance layer is secure, observable, and attributable. This enables agentic execution at scale without losing brand control.
Adobe's framework relies heavily on open standards to achieve this, specifically utilizing:
Adobe recognizes that there is no one-size-fits-all approach to AI deployment. To integrate these capabilities into daily operations, businesses can adopt one of three flexible orchestration models, depending on their team's needs and technical maturity:
Choosing the right model depends heavily on your team's technical maturity, existing tool ecosystem, and governance requirements.
For agents to act on behalf of a business, they also need a rules-based engine that can evaluate options and make policy-compliant decisions in real time.
In practice: Adobe Journey Optimizer (AJO) acts as the decisioning engine, using new semantic AI decisioning to autonomously recommend the next best journey or step.
Finally, this culminates in agentic experiences, where the end customer interacts with a highly personalized, agent-led journey.
In practice: Bringing the entire stack together is the Adobe CX Enterprise Coworker. A human marketer can set an objective — like a 3% lift in cross-sell performance — and the Coworker will autonomously coordinate across Real-Time CDP, Customer Journey Analytics, and Journey Optimizer to assemble the audience, pull creative assets, and build a journey plan, waiting only for final human sign-off before executing
The human-in-the-loop model is also what makes this approach viable for enterprise marketing teams today. Full autonomy is the aspiration, but deliberate oversight is what builds the organizational trust to get there
Deploying this type of stack requires a measured and realistic strategy so organizations can avoid "big bang" overhauls. When considering implementing an agentic stack, I recommend a POC (proof of concept) approach. By testing specific vertical slices of the capability stack incrementally — from data readiness up through governance to final experiences — teams can demonstrate tangible value early on and clearly define the requirements for future capabilities before rolling them out at scale. This approach also enables the team to comfortably execute and work alongside agentic AI.
Get ready for the final installment of our Adobe Summit 2026 series! In our upcoming post, I’ll explore how teams can move beyond basic reporting to achieve customer intelligence and cross-channel decisioning within the Adobe solution suite.
And if your organization is working through any of the layers covered here — from data readiness to orchestration model selection — we'd love to talk through what that looks like in practice.