Adobe’s new agentic AI platform for enterprises

For the past decade, Adobe has been the quiet giant of the marketing technology landscape. While the world obsessed over consumer-facing AI – chatbots, image generators, voice assistants – Adobe built the plumbing that powered the modern customer experience. Its Creative Cloud gave designers their tools. Its Experience Cloud gave marketers their data. Its Document Cloud gave enterprises their workflows. And Adobe sat comfortably in the middle, connecting brand to customer, taking a small toll on every interaction.

Then the agents arrived. And the middle started to disappear.

Over the last 18 months, a new generation of AI-native tools has threatened to bypass Adobe’s carefully constructed ecosystem. Figma launched Figma Agents , allowing designers to generate prototypes from natural language. Canva unveiled Canva Agents , automating social media content creation. And most ominously, Anthropic introduced Claude Design – a model that can generate production-ready design assets directly, without any intermediate tool at all.

The message was clear: why route your creative workflow through Adobe when the AI can just do the work? Why pay for a marketing orchestration platform when an agent can coordinate campaigns directly? The design world was moving toward agentic workflows, and Adobe was at risk of being the middleman that got cut out.

Today, at its annual Adobe Summit in Las Vegas, the company fired back.

Adobe CX Enterprise is the company’s answer to the agentic revolution – a new platform designed to help businesses coordinate marketing, content, and customer interactions through networks of AI agents. Unlike point solutions that replace specific tools, CX Enterprise aims to be the orchestration layer that sits above them. It weaves together three pillars under a single agentic roof: brand visibility (understanding how the brand is perceived), content supply chain (producing assets at scale), and customer engagement (delivering personalized experiences).

The centerpiece is the CX Enterprise Coworker – an agentic assistant that assembles the correct agents and tools based on a user’s goal, creates a plan, and executes multi-step actions across Adobe and third-party applications. A marketing team that wants to “increase cross-sell performance by three percent” can describe that goal to the Coworker, which then coordinates audience segmentation, creative asset generation, campaign deployment, and performance monitoring – all without the team moving between tools.

Adobe is also launching an agent skills catalog – reusable, customizable workflows that enterprises can deploy across their agentic infrastructure. And in a striking departure from its historically closed ecosystem, Adobe is announcing deep interoperability with every major AI platform: AWS, Anthropic, Google Cloud, IBM, Microsoft, NVIDIA, and OpenAI. Adobe’s Marketing Agent will plug directly into ChatGPT Enterprise, Claude Enterprise, Gemini Enterprise, and Microsoft 365 Copilot.

“Adobe CX Enterprise enables businesses to scale agentic AI with a fully customizable solution that is tailored to the needs of their organization,” said Anil Chakravarthy, President of Adobe’s Customer Experience Orchestration business. “This end-to-end solution fits naturally into any environment, built to work alongside tools across leading AI platforms with seamless interoperability.”

The message to customers is clear: you don’t have to choose between Adobe and the new AI agents. You can have both. Adobe will be the conductor, not the instrument. The question is whether the market still needs a conductor when the instruments can play themselves.

Part I: The Three Pillars – Brand, Content, and Engagement

CX Enterprise is not a single product but a system – an end-to-end agentic AI platform that spans the entire customer lifecycle. It is built on three pillars, each representing a domain where Adobe has deep historical expertise.

Pillar 1: Brand Visibility

The first pillar is about understanding and protecting the brand. Adobe has introduced Brand Intelligence, described as a “continuously learning reasoning engine” that captures evolving brand signals from across the web, social media, customer feedback, and internal communications. Agents in this pillar monitor brand perception, detect emerging trends, and flag potential reputational risks.

For example, a Brand Visibility agent might notice that customers on social media are associating the brand with a new attribute (e.g., “sustainable” or “expensive”) and alert the marketing team. Another agent might compare generated content against brand guidelines, flagging assets that deviate from approved colors, fonts, or messaging.

“Brand consistency is the first thing that breaks when you scale content production,” said Dr. Elena Vasquez, a marketing technology analyst. “Adobe is betting that enterprises will pay a premium for agents that enforce brand guardrails automatically, without human review.”

Pillar 2: Content Supply Chain

The second pillar addresses the production and distribution of marketing assets. Adobe’s existing Content Supply Chain tools – which include asset management, workflow automation, and creative collaboration – are now agentified. Agents can:

Generate on-brand creative assets using Adobe Firefly (the company’s generative AI suite)

Resize and reformat content for different channels

Translate copy into multiple languages

Archive obsolete assets and surface underutilized ones

The content supply chain agents integrate with Adobe’s Creative Cloud apps (Photoshop, Illustrator, Premiere) but also with third-party tools via MCP (Model Context Protocol) endpoints. A brand could have an agent that generates a social media post in Canva, checks it against brand guidelines via Adobe’s Brand Intelligence, and then schedules it in Sprout Social – all without human intervention.

Pillar 3: Customer Engagement

The third pillar is the most directly competitive with other AI platforms. Engagement Intelligence is a decisioning engine optimized for customer lifetime value. It ingests real-time data from Adobe Experience Platform (AEP) – which powers over one trillion customer experiences annually – and uses that data to inform agentic actions.

A Customer Engagement agent might:

Identify a segment of customers at risk of churn

Generate personalized retention offers

Deploy those offers via email, SMS, or push notification

Monitor responses and iterate on the offer in real-time

The key differentiator is Adobe’s data footprint. AEP is widely adopted among large enterprises, and it contains decades of customer interaction data across channels. Agents built on AEP have access to context that agents built on general-purpose models do not.

“Adobe’s advantage is not its AI,” said Marcus Wei, an enterprise software strategist. “It’s its data. They know more about how enterprises actually market to customers than any other company. That data is the moat. The agents are just the drawbridge.”

Part II: The CX Enterprise Coworker – Orchestration as a Service

The most visible component of CX Enterprise is the Coworker – an agentic assistant that abstracts away the complexity of coordinating multiple agents and tools.

Unlike a chatbot that answers questions or a copilot that suggests actions, the Coworker is designed to execute multi-step objectives. A user describes a goal in natural language – for example, “Launch a back-to-school campaign targeting parents of teenagers, with a 10% discount on backpacks, and measure conversion lift” – and the Coworker:

Assembles the right agents and tools (audience segmentation, creative generation, campaign deployment, analytics)

Creates a plan (audience first, then creative, then deployment, then measurement)

Requests approval at key decision points (e.g., “Here is the audience segment of 2.4 million users. Approve?”)

Executes the approved plan

Monitors results and reports back

The Coworker is built on Adobe’s Agent Orchestrator – a reasoning engine that understands the capabilities of available agents and can decompose high-level goals into concrete actions. It is, in effect, a meta-agent: an agent that coordinates other agents.

“The Coworker is the most interesting part of the announcement,” said Sarah Jenkins, a customer experience consultant. “It acknowledges a reality that most enterprises are facing: they have too many agents, too many tools, and no way to coordinate them. The Coworker is Adobe’s bet that they can be the universal remote control for the agentic enterprise.”

The Coworker will be generally available “in the coming months,” according to Adobe. Pricing has not been announced, but it will likely be sold as an add-on to existing Adobe Experience Cloud subscriptions – a way for Adobe to monetize its orchestration layer without forcing customers to abandon their existing investments.

Part III: The Open Ecosystem – Adobe’s Unlikely Alliance

The most surprising aspect of CX Enterprise is how open it is. Adobe, historically a company that preferred customers to live entirely within its walled garden, is now announcing deep interoperability with every major AI competitor.

The Adobe Marketing Agent – a specialized agent for campaign execution – will be pluggable into:

Amazon Quick (AWS’s business intelligence tool)

Anthropic Claude Enterprise

ChatGPT Enterprise

Gemini Enterprise (Google Cloud)

IBM watsonx Orchestrate

Microsoft 365 Copilot

This means that a user working in Microsoft Teams could summon the Adobe Marketing Agent via Copilot, ask it to generate a campaign report, and receive an answer without ever opening an Adobe application. The underlying workflows – data retrieval, analysis, formatting – happen in Adobe’s cloud, but the user interface is wherever the user already is.

“This is Adobe admitting that they cannot win the UI war,” said Wei. “They are not going to replace Slack or Teams or Copilot. So instead of fighting those surfaces, they are embedding into them. It is a defensive move, but it is a smart one.”

Adobe is also embracing MCP (Model Context Protocol) – the open standard originally developed by Anthropic – as the connective tissue for its agentic ecosystem. MCP servers will enable developers to pipe data from Adobe’s platforms into any MCP-compatible agent, and vice versa. An enterprise could build a custom agent in Claude that queries Adobe’s Engagement Intelligence, then uses that data to inform a campaign in a non-Adobe tool.

“The MCP adoption is significant,” said Dr. Alex Chen, an AI infrastructure researcher. “It means Adobe is not trying to own the protocol. They are contributing to an open standard. That lowers the barrier for customers to adopt CX Enterprise, because they know they won’t be locked in.”

Whether this openness is genuine or tactical remains to be seen. Adobe has a long history of “open” integrations that are, in practice, shallow and difficult to maintain. But the press release is unusually explicit about the company’s commitment to “a composable architecture that works with various technology stacks.” Investors and customers will be watching closely to see if Adobe delivers.

Part IV: The Agent Skills Catalog – Reusable Workflows

One of the more practical announcements in CX Enterprise is the agent skills catalog – a library of reusable, customizable instructions that enterprises can deploy across agents.

A “skill” is essentially a packaged workflow: a set of instructions, data sources, and tool integrations that accomplish a specific task. For example, a “Customer Churn Risk Assessment” skill might include:

Instructions: “Analyze last 90 days of usage data, support ticket history, and payment status. Flag any account with >2 support tickets and decreasing usage.”

Data sources: AEP segments, Zendesk tickets, Stripe payment data

Integrations: Slack (to send alerts), Salesforce (to create tasks)

Once a skill is created, it can be reused across multiple agents and workflows. A marketing agent might use the churn skill to identify at-risk customers for a retention campaign. A sales agent might use the same skill to prioritize outreach. An operations agent might use it to trigger proactive support interventions.

“The skills catalog is Adobe’s answer to the ‘shadow workflow’ problem,” said Jenkins. “Right now, every team builds its own automations. They don’t share. They don’t reuse. The skills catalog is a central repository for best practices – if Adobe can get teams to actually use it.”

Enterprises can customize skills with their own context – adding brand guidelines, compliance rules, or business logic. Adobe will also provide a set of pre-built skills based on its domain expertise in marketing, creative, and customer experience. Over time, the company hopes to build a marketplace of skills, similar to what the GPT Store attempted (and largely failed) to achieve.

“The difference is that Adobe’s skills are not standalone products,” noted Vasquez. “They are building blocks for enterprise workflows. That is a much more defensible market than consumer GPTs. Enterprises will pay for well-documented, well-supported, reliable skills. Adobe has the trust to sell them.”

Part V: The Competitive Landscape – Figma, Canva, and the Labs

Adobe’s move into agentic orchestration does not happen in a vacuum. The entire design and marketing technology landscape is shifting toward agents, and Adobe is now one of several players jockeying for position.

Figma Agents: Launched in late 2025, Figma Agents allow designers to generate prototypes, components, and design systems from natural language. The agents are deeply integrated into Figma’s collaborative design environment. For teams already using Figma for design, the agents eliminate the need to export assets to other tools – including Adobe’s.

Canva Agents: Canva, the democratized design platform, has been quietly building agentic capabilities for content creation and social media scheduling. Its agents are less powerful than Figma’s for professional design, but they are far more accessible to non-designers. For small and medium businesses, Canva Agents may be sufficient – and much cheaper than Adobe.

Claude Design: The most disruptive threat comes from Anthropic. Claude Design – announced in March 2026 – can generate production-ready design assets directly from text prompts, including layouts, color schemes, typography, and even simple animations. It does not require any intermediate design tool. A user can describe a landing page, and Claude Design outputs HTML, CSS, and assets. The middleman – whether Figma, Canva, or Adobe – is completely bypassed.

“Claude Design is the existential threat,” said Chen. “It doesn’t need Adobe. It doesn’t need Figma. It just needs the prompt. If that capability continues to improve, the entire market for design tools collapses. Adobe’s response is to become the platform that coordinates Claude Design alongside other tools. That is a smart pivot, but it is a pivot from a position of weakness.”

Adobe’s counterargument is that enterprises will never trust a single model – no matter how powerful – to handle their entire marketing workflow. They will always need governance, brand consistency, data integration, and auditability. Those are Adobe’s strengths. The agents are just the execution layer; Adobe provides the safety rails.

“The bigger threat is the labs cutting out the middleman,” Chakravarthy acknowledged in a briefing with analysts. “Launches like Claude Design and every subsequent improvement will make legacy orchestration paths more difficult. That is why we are building CX Enterprise – not to compete with Claude, but to orchestrate with Claude. We want to be the layer that enterprises trust to govern all agents, regardless of who built them.”

Part VI: Why It Matters – The Orchestration Layer as a Moat

The most important question for Adobe investors and customers is whether CX Enterprise can succeed where previous “orchestration” platforms have failed. The history of enterprise software is littered with companies that tried to be the middleware layer, only to be bypassed as point solutions became more powerful.

Adobe’s bet is that the agentic future will be heterogeneous – not a single model or tool that does everything, but a swarm of specialized agents, each best-in-class for its domain. A content generation agent from Anthropic. A data analysis agent from OpenAI. A creative asset agent from Adobe. A campaign execution agent from Salesforce. Enterprises will need a way to coordinate these agents, to enforce governance across them, and to provide a unified interface for human users.

That is the role Adobe is claiming: the orchestration layer for the agentic enterprise.

“Adobe’s historical advantage has been the breadth of its suite,” said Wei. “Creative Cloud, Document Cloud, Experience Cloud – they have something for every part of the customer journey. In the agentic era, that breadth becomes even more valuable. Because agents need context. And Adobe has more context than anyone.”

But context is not destiny. The orchestration layer is valuable only if enterprises cannot easily bypass it. And today, they can. A marketer could use Claude Design to generate creative assets, GPT-5.4 to analyze customer data, and a custom script to coordinate them – all without touching Adobe. The friction is low, and the cost is falling.

Adobe’s answer is to make the friction of bypassing Adobe higher than the friction of using CX Enterprise. That means deep integrations, pre-built skills, a robust governance layer, and – most importantly – the ability to bring an enterprise’s own data into the orchestration workflow without moving it. Adobe Experience Platform, with its 1 trillion annual customer experiences, is the anchor.

“If Adobe can make CX Enterprise the easiest way to deploy agentic workflows on top of AEP, they have a business,” concluded Jenkins. “If they can’t, they will be disintermediated. It is that simple. The next 18 months will tell us which way it goes.”

Conclusion: The Middleman Fights Back

Adobe CX Enterprise is not a product launch. It is a statement of intent. Adobe is acknowledging that the agentic revolution is real, that it threatens the company’s core business, and that the only way to survive is to embrace it – not by building better agents than Anthropic or OpenAI, but by becoming the platform that orchestrates all agents.

The strategy is risky. It requires customers to trust Adobe as an impartial coordinator, even as Adobe also sells its own agents. It requires deep technical investment in interoperability, an area where Adobe has historically underinvested. And it requires a cultural shift inside Adobe – from a company that builds tools to a company that connects them.

But the alternative is worse. If Adobe does nothing, the agents will eat its lunch. Figma, Canva, and the AI labs are already circling. Adobe is fighting back.

For the 20,000 global brands that have built their businesses on Adobe, the message is clear: you don’t have to choose between Adobe and the new AI. You can have both. Adobe will be the conductor. The agents will be the instruments. And together, they will play the music of modern marketing.

Whether the market still wants a conductor is a question only the next few years will answer.