Adobe just retired the Experience Cloud brand. At Adobe Summit in Las Vegas on April 20, 2026, the company unveiled Adobe CX Enterprise, an end-to-end agentic AI platform that folds creative, marketing, data, and commerce capabilities into a single agent-callable surface. The headline product is CX Enterprise Coworker—a persistent, self-learning agent that can carry a business objective across weeks of campaign work without losing context. The pricing model shifts from seats and products to credits tied to agent actions. The customer list Adobe claims already entitled to this model: 1,770-plus enterprises.

Why this matters now: Adobe is the second major SaaS vendor in five days—after Salesforce's Headless 360 on April 15—to rebuild its platform around agents instead of interfaces. For CIOs, CMOs, and CFOs still treating 2026 AI as a pilot budget, this is the signal that the seat-licensed SaaS era is ending faster than most IT plans assume. The question is no longer whether your martech stack becomes agentic. It is whether your commercial model, your governance posture, and your integration architecture are ready when it does.

What CX Enterprise Actually Ships

Adobe reorganized its product line around three pillars instead of product brands. CX Enterprise is now structured as Brand Visibility (how your brand shows up across owned, earned, and AI-generated surfaces), Customer Engagement (personalization, journeys, decisioning), and Content Supply Chain (asset production, governance, activation). That map replaces a decade of Marketo, Workfront, Target, and Campaign brand sprawl. For practitioners, the apps still exist. For buyers, the SKU conversation changes.

The runtime is Adobe Experience Platform (AEP) Agent Orchestrator plus a new Adobe AI Platform. AEP already powers over a trillion customer experiences annually across roughly 20,000 enterprises. Adobe is now layering two intelligence systems on top: Brand Intelligence, a fine-tuned vision-language model that learns from a customer's own asset annotations and feedback, and Engagement Intelligence, which scores next-best actions across a customer's entire lifetime history rather than a single session. Neither system requires customers to ship data to a third-party foundation model—Adobe sits between the enterprise's first-party data and the chosen LLM.

Over ten purpose-built agents are in production today. The current slate covers site optimization, audience creation, journey orchestration, experimentation, content optimization, LLM optimization (specifically for how brands show up inside AI answer engines), and data insights. Each agent is callable directly or chainable by the Coworker layer.

CX Enterprise Coworker is the architectural bet. Unlike point-task agents that execute single prompts, the Coworker operates persistently—with memory that spans weeks and the ability to translate a business objective like "reduce churn among mid-market accounts next quarter" into a sequence of audience builds, creative refreshes, journey edits, and measurement loops. Adobe VP Sundeep Parsa described the model in the launch briefing as a quarterback "guided by a coach on the sidelines," where the coach is a human marketer approving plays before they ship. General availability is slated for the coming months; today, the Coworker is in production beta with named accounts.

Model Context Protocol is Adobe's interoperability story. CX Enterprise exposes MCP endpoints across its products and publishes reference architectures for Claude, ChatGPT Enterprise, Gemini Enterprise, Microsoft Copilot, and Nvidia NemoClaw. Partnerships with AWS, Anthropic, Google Cloud, IBM, Microsoft, Nvidia, and OpenAI are extended, not exclusive. Adobe is explicit that it is a system of engagement and content, not a system of foundation models—customers pick the brain, Adobe provides the hands.

For CIOs: Four Things You Inherit Whether You Want Them or Not

Adobe's pitch to your technical team is that CX Enterprise hands agents four layers most homegrown martech stacks never assemble cleanly. Strip the marketing language and these are worth evaluating against your roadmap.

1. Customer data semantics. AEP is already the Real-Time CDP for many large enterprises. Exposing it as an MCP-callable surface means an agent doesn't need a bespoke RAG pipeline to ground on your customer record. If you have already paid the integration cost of putting identity, consent, and behavioral data in AEP, your agents inherit that grounding for free.

2. Governance and brand policy. Brand Intelligence is, functionally, a machine-readable brand book. It enforces tone, visual identity, and claims language on agent-generated creative before it goes to approval. This is the layer most "let's vibe-code our marketing" prototypes never build, and it is where regulated industries—financial services, healthcare, pharma—have to live.

3. Orchestration across channels. The Coworker plus the Engagement Intelligence system means a single agent can coordinate email, web, mobile, paid, and in-product experiences against one customer identity. Homegrown stacks tend to get stuck at two or three channels before the integration debt swamps the team.

4. Measurement. Adobe's analytics heritage is still a differentiator. CX Enterprise agents write back into AEP and Customer Journey Analytics, so you get a closed measurement loop instead of the "agent did a thing, we'll report on it next quarter" pattern that plagues first-generation agent deployments.

The honest CIO read: these four layers are valuable in direct proportion to how deep Adobe already sits in your stack. Heavy AEP customers get a step-change in agent readiness. Companies running Adobe as one of three or four competing martech stacks will find the lock-in implications more uncomfortable than the productivity gains. Plan the conversation accordingly.

For CTOs: Where the Governance Work Actually Lives

The Coworker's multi-week memory is the feature that matters, and it is also the feature that will keep your security team up at night. A persistent agent that retains context across campaigns is, by definition, a persistent repository of customer data, brand decisions, and internal strategy. Adobe's answer is that Brand Intelligence and Engagement Intelligence sit inside AEP's existing permission model, that the Coworker requires human approval before anything customer-facing ships, and that observability is part of the platform—not an add-on.

Take that at architecture-review rigor, not keynote rigor. Specifically: ask your Adobe AE to walk you through the Coworker's memory isolation model across business units, its data-retention controls, and its behavior when a source data contract changes mid-campaign. Ask how the approval gate handles sub-actions the Coworker takes before the final creative is produced (e.g., querying competitor data, testing prompt variations). The answers will tell you how much governance work lives in Adobe versus how much still lives in your CASB, DLP, and IAM stack.

The MCP surface is the other place to probe. Exposing AEP as MCP endpoints is powerful; it also means any agent with the right credentials can call your customer data. Make sure the scoping of MCP tokens—per agent, per business unit, per use case—is as granular as your existing API governance. If Adobe's scoping is coarser than your internal standard, you have work to do before production.

Don't skip the evaluation piece. The new practitioner dashboards tracking agent usage are in beta. Treat this as a 2027-complete feature, not a 2026-complete one. For regulated use cases, budget engineering time to build your own eval rubrics and session-trace monitoring on top of what Adobe ships.

For CFOs and CMOs: The Pricing Model Is the Business Model

Adobe is moving to credit-based consumption for AI agents plus value-based metrics for new AI-first apps. The public number is the 1,770-plus customers already entitled to use the new model, with detailed practitioner dashboards tracking usage patterns in beta. What is not on the price sheet yet: the per-credit cost, the agent-action catalog (how many credits a Coworker burns to orchestrate a quarterly campaign), or the overage posture.

That opacity is the CFO risk. Consumption pricing is CFO-friendly until your agents start running at 10x the volume of human users—which is exactly the point of agents. Without a hard cap and a transparent action catalog, your 2026 Adobe bill becomes a function of how aggressively your marketing team deploys the Coworker, not a line item you can forecast.

Three questions to put to your Adobe AE in writing:

• What is our worst-case credit burn if the Coworker runs at 5x our forecast usage in month one, and how are overages priced?
• Which existing Experience Cloud SKUs get folded into CX Enterprise credits, and what is the true-up math at renewal?
• What is the minimum credit commitment required to access Coworker GA, and does an early commit lock in pricing before Adobe formalizes the catalog?

For CMOs, the business case is tighter than the IT read. Adobe's early-production claims include "massive" reductions in creative troubleshooting (days and weeks compressed to hours) and measurable gains in campaign iteration speed. Both are vendor-reported and pre-GA. Treat them as upper-bound hypotheses. The CMO KPI that will actually move first is time-to-launch for a targeted campaign—if the Coworker can genuinely compress that from a six-week cycle to a one-week cycle on a real program of yours, the ROI case writes itself. If it cannot, no amount of agent branding will change your CAC.

Competitive Read: Adobe Against Salesforce, Microsoft, and the Point Players

Salesforce is the mirror competitor. Headless 360 (April 15) and CX Enterprise (April 20) are philosophically identical bets: expose the platform as agent-callable infrastructure, move pricing toward consumption, lean on MCP for interoperability. The practical split is still historic: Salesforce owns the system-of-record customer data and the selling motion; Adobe owns the content supply chain and the brand layer. Enterprises running both will spend 2026 deciding which platform's agents orchestrate the other's data.

Microsoft is the platform competitor. Copilot Studio plus Foundry plus Dynamics 365 gives Microsoft an agent runtime, an eval layer, and a customer-data backplane—but Microsoft's marketing and content depth is thinner than Adobe's. Expect Microsoft to push Copilot-native customer journeys, and expect Adobe to counter by being Microsoft's preferred creative and content layer.

Anthropic and OpenAI are the disruptors. On the same day Adobe launched CX Enterprise, Anthropic unveiled Claude Design, a visual creation feature that lets users generate prototypes and slide decks directly from Claude. This is still a demo-level capability, not an enterprise platform—but it is the direction every foundation-model vendor is moving. Adobe's five-year bet is that brand safety, governance, and enterprise data integration are moats that generic AI tools will not cross quickly. That bet is credible, but it requires Adobe to keep shipping governance features faster than Anthropic and OpenAI can ship enterprise packaging.

The SaaS-stock context is real. Adobe shares are down roughly 30% year-to-date as investors price in disruption from AI-first alternatives. The stock popped 2.2% on launch day. The longer-term read: Adobe has to prove that "agentic SaaS" is a premium tier, not a commodity layer over foundation models. CX Enterprise is that proof-of-concept.

The 90-Day Action Plan

1. Get a CX Enterprise Coworker sandbox stood up in two weeks. Pick one business unit, one campaign type, and one measurable KPI (campaign launch time, conversion lift, or creative iteration velocity). Run the Coworker against a real but non-production workload.

2. Map your MCP and identity boundaries. Before any production agent touches customer data, document how MCP tokens are scoped, how Adobe's permission model interacts with your IAM, and where the Coworker's memory is physically stored.

3. Pressure-test the pricing. Model the credit burn for your top three marketing workflows at current volume, then at 3x and 10x. Demand an overage cap in writing before you sign.

4. Re-examine your martech stack. If CX Enterprise absorbs three or four existing line items (Marketo, Workfront, Target, a CDP), model the consolidation savings. If it displaces a competing stack (Salesforce Marketing Cloud, Braze, a homegrown orchestration layer), map the migration risk honestly.

5. Loop in the CMO and the General Counsel now. This is not an IT decision. Brand agents change how compliant content is produced and approved; legal needs to be in the room before the pilot, not after.

6. Decide your agent-runtime posture. Are you going to run a multi-vendor agent strategy (Adobe for marketing agents, Salesforce for sales agents, Microsoft for productivity agents) or consolidate on one runtime that calls all of them via MCP? That decision dictates which CX Enterprise features matter most.

The Bottom Line

Adobe's CX Enterprise is the second unmistakable signal in a week that enterprise SaaS is being re-architected around AI agents. The platform is well-architected, the MCP and multi-model commitments are real, and the Coworker concept is more honest about persistent agent work than most vendor pitches this year. The risks—credit-pricing opacity, vendor concentration, the still-early maturity of the Coworker and its observability—are equally real.

The most important thing Adobe did is normalize two things the broader industry was resisting: persistent agents with multi-week memory and value-based pricing tied to agent actions, not seats. Every major martech and CX vendor will have a version of this by end of year. Your 2026 martech roadmap has to assume both, and your governance and procurement posture has to be ready for both.

Don't sign the enterprise agreement on Adobe Summit keynote energy. Run the sandbox, write the evals, model the worst-case credit bill, and decide whether inheriting Adobe's brand-data-governance-measurement layers is worth the consolidation on a single vendor. For heavy Adobe shops, the answer is likely yes. For everyone else, MCP standardization is still giving you optionality—use it while you have it.

Want to calculate your own AI ROI? Try our AI ROI Calculator — takes 60 seconds and shows projected savings, payback period, and 3-year ROI.

Continue Reading

• Salesforce Headless 360: The UI Becomes Optional
• Salesforce Aela Flat-Fee Agentic Pricing: What CFOs Should Know
• Anthropic Claude Design: The Visual Tool Coming For Enterprise Creative

Sources

• Adobe: Adobe Redefines Customer Experience Orchestration Vision in the Agentic AI Era with Introduction of CX Enterprise
• Adobe: Adobe Unveils CX Enterprise Coworker
• CIO: Adobe bets on agentic AI to rewrite SaaS for customer experience
• MarTech: Adobe rebrands Experience Cloud as 'CX Enterprise,' goes all-in on AI agents
• SiliconANGLE: Adobe deploys AI agents across its customer experience tools
• Allwork.Space: Adobe Launches New AI Suite As Anthropic And OpenAI Intensify Enterprise Software Competition