Table of Contents
- What Adobe CX Enterprise Coworker Is
- The November 4 Announcement: Five Platforms at Once
- Where Adobe CX Enterprise Coworker Is Available Now
- What the Coworker Actually Does for Marketers
- How MCP and A2A Make It Work
- The New MCP Connectors
- Who Can Actually Use It — and What You Need First
- What Adobe Did Not Announce
- Why This Matters: Agents Go Where the Work Already Happens
- Adobe CX Enterprise Coworker vs the Alternatives
- What to Watch Next
- Frequently Asked Questions
- The Bottom Line
Adobe CX Enterprise Coworker has left Adobe’s own walls. On 4 November 2026, Adobe announced that its agentic teammate for customer experience work now runs inside ChatGPT, Claude, Databricks Genie, Gemini for business and Microsoft Copilot — the five AI assistants where marketers and analysts already spend their working day. Instead of opening yet another dashboard, a marketer can now ask the assistant they already use for a campaign forecast, an audience analysis or a journey check, and the answer comes back grounded in Adobe’s customer data rather than a generic guess.
That shift matters more than it first looks. For years, enterprise AI lived in the vendor’s own product: you bought the platform, you learned its interface, and the agent worked only there. Adobe CX Enterprise Coworker flips the model. The agent travels to the user. The announcement is also a bet on open standards — the Model Context Protocol (MCP) and Agent2Agent (A2A) — over walled gardens, and it lands in a week when every major platform is racing to put agents where people already work.
This guide explains what Adobe CX Enterprise Coworker is, exactly where it is available now, what it can and cannot do inside each platform, what your organisation needs before switching it on, and the questions Adobe has not answered yet — including pricing, which the company did not disclose at all.

What Adobe CX Enterprise Coworker Is
Adobe CX Enterprise Coworker is an end-to-end agentic AI system that helps enterprises plan, run and improve the whole customer lifecycle — from finding and engaging prospects, through converting them, to growing their lifetime value. Adobe describes it as an “agentic teammate” for customer experience orchestration, which is a deliberately human word: the system is meant to work alongside a marketing team, taking on multi-step work rather than answering one question at a time.
The product is not new. Adobe CX Enterprise Coworker reached general availability in June 2026, so it had already been running inside Adobe’s own Customer Experience (CX) Enterprise environment for five months before this announcement. What changed on 4 November is distribution. The same agent is now reachable from outside Adobe’s interface, inside the general-purpose AI platforms that marketers, analysts and executives already open every morning.
Adobe says more than 20,000 brands use its tools to drive performance and protect brand integrity across the content supply chain, customer engagement and brand visibility. That installed base is the reason the expansion matters: if even a fraction of those organisations switch on the new access, Adobe’s agent suddenly appears in millions of daily ChatGPT, Claude, Gemini and Copilot sessions. The company is not asking users to come to Adobe. It is placing its agent where its users already are.
Under the surface, Adobe CX Enterprise Coworker is built on two open standards. The Model Context Protocol (MCP) lets the agent connect to external tools and data sources through pre-built servers, and Agent2Agent (A2A) lets agents from different vendors talk to each other and hand work back and forth. Adobe leans on both standards heavily in its announcement, and that choice is part of the story: the company is positioning open plumbing, rather than a closed Adobe-only world, as the way enterprise agents should connect.
Human oversight is built into the design, at least as Adobe describes it. The agent powers autonomous Customer Experience Orchestration workflows “with human oversight”, which means people stay in the loop on the work it plans and executes. How that oversight works in practice for Adobe CX Enterprise Coworker — approval steps, audit trails, who can override what — is one of the details Adobe has not spelled out for the new third-party access, and it is worth asking about before any team switches it on.
The November 4 Announcement: Five Platforms at Once
Adobe announced the expansion on 4 November 2026 from San Jose, California, in a press release carried on the Adobe newsroom and BusinessWire. The headline was simple: Adobe CX Enterprise Coworker is now available within leading AI platforms, with three more to follow — a single-day distribution leap for Adobe CX Enterprise Coworker. The release named the platforms precisely, and the precision matters, because the availability picture is split into two groups — five that work now, and three that are promised without a date.
The five live platforms are:
- ChatGPT — available as a plugin inside OpenAI’s assistant
- Claude — available inside Anthropic’s assistant
- Databricks Genie — available inside Databricks’ AI/BI assistant
- Gemini for business — available inside Google’s business Gemini offering
- Microsoft Copilot — available inside Microsoft’s Copilot
Three more are expected soon, with no timeline given: Amazon Quick, IBM and Snowflake. Adobe used the words “expected soon” and gave no date, so treat those three as a roadmap, not a launch. If your team runs on one of them, the honest position today is that you are waiting.
The announcement also carried quotes from the partners, which tell you how each side is framing the move. Anjul Bhambhri, Adobe’s Senior Vice President of Engineering for Customer Experience Orchestration, said marketers “often use AI tools in isolation, without the organisational context and domain intelligence needed to personalise every stage of the customer lifecycle” — and that CX Enterprise Coworker connects those tools and workflows using brand, customer and channel intelligence. The pitch, in other words, is context: the agent is useful because it knows your brand, your customers and your channels, not because it is a smarter chatbot.
Google’s Rao Surapaneni, Vice President of Gemini Agent Governance at Cloud AI, called the integration a way for organisations to “securely harness their customer data and brand intelligence right where they already work”, and described the agent-to-agent approach as “the next frontier of productivity and creative execution”. Microsoft’s Rob Howard, Corporate Vice President of Copilot Platform and Extensibility, said the two companies are working to help enterprises “move from insight to action” by connecting intelligence across customer data, content, journeys and enterprise systems.
Read together, the quotes make the strategy clear. Adobe is selling reach and context; Google and Microsoft are selling the idea that their assistants become more useful when an enterprise agent lives inside them. Nobody in the announcement talked about replacing marketers. The language is consistently about orchestration, speed and personalisation at scale.
Where Adobe CX Enterprise Coworker Is Available Now
Not every platform gets the same thing. Adobe’s announcement says Adobe CX Enterprise Coworker is “available within” all five platforms, but the details of Adobe CX Enterprise Coworker differ in ways that matter if you are choosing where to use it.
ChatGPT gets the agent as a plugin. That is the most familiar model for ChatGPT users: you add the plugin, and the assistant can reach Adobe’s agent when a question needs customer data. Claude, Databricks Genie, Gemini for business and Microsoft Copilot get the agent as an integration inside the assistant itself. Adobe has not published a feature-by-feature comparison across the five, so do not assume identical behaviour in each one. One difference is already confirmed: Firefly Image Generation is unavailable through the Claude integration, according to Adobe’s own setup documentation summarised in independent coverage. If image generation is part of your workflow, Claude is not the door to use for it today.
Access to Adobe CX Enterprise Coworker through these platforms follows the permissions tied to a user’s Adobe identity, and the integrations return read-only insights rather than modifying customer data, according to Adobe’s setup guidance. That is an important guardrail: asking Copilot or Claude a question about your campaigns will not rewrite your audience segments or push a change live. It will show you analysis and recommendations drawn from your Adobe data. The trade-off is clear — safety first, action second — and it is the opposite of the fully autonomous agent some headlines imply.
Usage of Adobe CX Enterprise Coworker also draws on a shared AI-credit pool — the same pool that other Coworker features use — so heavy use inside ChatGPT or Copilot consumes the same credits as use inside Adobe’s own interface. Organisations with capped credit allowances should check their balance before rolling this out to a large team, because five new doors into the same pool can drain it faster than one.
There is also a practical point about who sees what. Because access follows Adobe permissions, a junior analyst and a marketing director asking the same question inside Gemini may get different answers — or different levels of detail — depending on what each of them is allowed to see in Adobe. That is good governance, but it means a team demo can look inconsistent if the demonstrator and the audience hold different permissions. Test with the accounts your team will actually use.
What Is Still Coming Soon
Adobe CX Enterprise Coworker is expected to arrive in Amazon Quick, IBM and Snowflake, but Adobe gave no date for any of the three. The Snowflake one is worth watching most closely, because Adobe also launched a pre-built MCP connector to Snowflake Cortex AI in the same announcement — so the plumbing between Adobe and Snowflake is already being built even while the full integration is pending. IBM’s inclusion is also notable, given IBM is one of the nine global systems integrators Adobe named as building on the agent; the line between partner and platform is blurring.
Until Adobe puts a date on these three, treat them as “announced, not available”. If a vendor or a reseller tells you otherwise, ask them to show you the live integration. The same discipline applies to any claim about feature parity: Adobe has not said the five live platforms offer identical capabilities, and the one confirmed gap — Firefly images missing from Claude — suggests they do not.
What the Coworker Actually Does for Marketers
So what does Adobe CX Enterprise Coworker actually do once it is inside your assistant of choice? Adobe groups the work of Adobe CX Enterprise Coworker into five areas, and each one maps to a job a marketing team already does by hand today.

Campaign Performance Analysis
Adobe CX Enterprise Coworker analyses campaign performance and spots emerging trends. In practice, that means asking a question in plain language — “which of our October campaigns underperformed, and why?” — and getting back an answer grounded in your own Adobe data instead of a generic explanation of how campaigns work. The agent can pull the numbers, compare campaigns against each other, and highlight the ones that moved. For a team running dozens of campaigns across channels, the value is not the chart; it is the triage. You start the morning knowing which three campaigns need attention, rather than opening five dashboards to find out.
Audience Forecasting
Audience forecasting is the second named capability of Adobe CX Enterprise Coworker, and it is the one most likely to change how planning meetings run. Instead of a marketer guessing how large an audience segment will be, or exporting a list to check, the agent forecasts the audience for a planned campaign using the customer data already in Adobe’s platform. A forecast that says “this segment will reach roughly 40,000 people, and it skews heavily toward repeat buyers” is a planning input a team can act on the same day. A forecast is still a forecast — it is a model output, not a guarantee — but a forecast grounded in your own first-party data is a very different thing from a generic benchmark pulled from the web.
Journey Optimisation and Conflict Resolution
The agent also optimises customer journeys and resolves conflicts between them. This is the least visible part of the product and possibly the most valuable. In a large organisation, several teams run journeys at once: onboarding emails, win-back offers, loyalty prompts, support follow-ups. Those journeys can collide — a customer can receive a discount offer and a premium upsell on the same day, or get the same message twice through two channels. Adobe says Adobe CX Enterprise Coworker spots those conflicts and helps resolve them, and keeps journeys coherent across channels. If you have ever been on the receiving end of a brand that emailed you three contradictory offers in one afternoon, you know exactly why this matters.
Planning, Project and Work Management
Finally, Adobe CX Enterprise Coworker helps with planning, project and work management for customer experience work. That is a broad label, and Adobe has not broken it into a feature list, but the shape of it is clear from the orchestration framing: the agent keeps track of the moving parts of a campaign — tasks, owners, deadlines, dependencies — so the team spends less time chasing status and more time doing the work. Used well, this is the difference between an agent that answers questions and an agent that runs the operation.
One honest note: everything above is Adobe’s description of what Adobe CX Enterprise Coworker does. The company has published no independent benchmark, no customer case study with measured results, and no named deployment for the new third-party access. The capabilities are plausible and the architecture is real, but the proof — a named brand saying “this saved us X hours” or “this lifted conversion by Y” — does not exist in public yet. Judge the product on its design and on your own trial, not on results Adobe has not shown you.
How MCP and A2A Make It Work
The technical heart of the announcement is not the list of platforms. It is the two open standards Adobe built Adobe CX Enterprise Coworker on: MCP and A2A.
MCP, the Model Context Protocol, is the standard that lets an AI assistant connect to tools and data sources through small pre-built servers. Think of it as a universal plug: instead of Adobe writing a separate custom integration for ChatGPT, another for Claude, another for Copilot, Adobe CX Enterprise Coworker exposes its capabilities through MCP servers that any compliant assistant can use. That is why five platforms could launch on the same day. The plumbing is shared.
A2A, Agent2Agent, is the complementary standard: it lets agents from different vendors communicate, delegate and hand work back and forth. When you ask Copilot a question that needs Adobe data, Copilot’s agent can pass the request to Adobe CX Enterprise Coworker, which does the work in Adobe’s systems and returns the result. Neither agent has to absorb the other. They cooperate as peers.
Why should a marketer care about plumbing standards? Because standards decide lock-in. If the agent only worked through Adobe’s own proprietary connections, switching assistants — or adding a sixth platform next year — would mean starting over. With MCP and A2A, the agent is portable by design. Adobe is betting that openness wins enterprise trust faster than a closed ecosystem does, and the bet is visible in the whole announcement of Adobe CX Enterprise Coworker: open standards named in the second paragraph, pre-built connectors for rival systems, and integrators invited to build on top.
This is also where the announcement connects to a wider shift. The same week, other enterprise tools were opening up through MCP too — Zoho Analytics added an MCP server that lets agents query business data, and Snowflake’s CoCo coworker took a similar agent-first approach to enterprise data. The pattern is unmistakable: 2026 is the year enterprise software stopped asking users to live inside its own window and started sending its agents out to where the users already work. Adobe’s move is the marketing-industry version of that shift.
The New MCP Connectors
Alongside the five platform integrations, Adobe launched a set of new pre-built MCP connectors for Adobe CX Enterprise Coworker — the pipes that let Adobe CX Enterprise Coworker reach beyond Adobe’s own systems. These let the agent reach into the partner systems a marketing team already uses, pull data out, and in some cases take action there — without custom integration work.
The connector list Adobe named is:
- Amazon Ads — advertising data and actions from Amazon’s ad platform
- Databricks Genie — the Databricks AI/BI assistant, also a host platform for the agent itself
- Demandbase — account-based marketing data
- Meta’s ads MCP server — campaign data from Meta’s advertising platform
- Microsoft Work IQ — Microsoft’s work-intelligence layer
- RainFocus — event marketing data
- Snowflake Cortex AI — Snowflake’s AI layer over enterprise data
The practical promise is a single request that crosses systems. A marketer should be able to ask one question — “how did our Meta and Amazon campaigns perform against the same audience last month, and what should we change?” — and get an answer that pulls from both ad platforms plus Adobe’s own customer data, in one place.
Adobe’s phrasing is that marketers can “connect the partner systems they already use without custom integration work, then use a single request to extract data and take actions across multiple systems”. Note the two verbs: extract data and take actions. The read-only guardrail discussed earlier applies to the assistant integrations; the connectors are framed more broadly, as a way to act across systems. Before relying on that, confirm with Adobe which actions are live through each connector and which need approval, because “take actions” without an approval step is a governance question, not just a feature.
The connector list also reveals Adobe’s alliance map. Meta and Amazon are the two largest advertising platforms in the world, and both appear here as data sources for Adobe’s agent rather than as competitors to it. Snowflake appears twice — as a connector today and as a host platform “expected soon” — which suggests the Adobe-Snowflake relationship is deepening in stages. Databricks appears in both roles too. The pattern is a hub: Adobe’s agent sits in the middle, and the pipes run out to wherever the data lives.
Who Can Actually Use It — and What You Need First
This is the section most coverage skips, and it is the one that decides whether your team can switch Adobe CX Enterprise Coworker on this week or not at all — because Adobe CX Enterprise Coworker is not a consumer app you can simply open and try. Access is not automatic, and it is not aimed at individual consumers. It is an enterprise feature with enterprise prerequisites.

According to Adobe’s setup guidance, an organisation needs all of the following before connecting the agent through a third-party assistant:
- Coworker provisioned for the organisation — the product must be set up and enabled by an Adobe administrator; there is no self-serve sign-up for an individual marketer.
- Adobe access for the user — each person needs their own Adobe identity, because permissions and data visibility follow that identity.
- Adobe Experience Platform — the data foundation the agent reads from must be in place.
- At least one qualifying Adobe solution — Real-Time CDP, Adobe Journey Optimizer, or Adobe Customer Journey Analytics.
In plain terms: if your company already runs Adobe Experience Platform plus one of those three products, and your administrator provisions Coworker, you are in scope. If your company uses Adobe’s creative tools but not its experience platform, this announcement does not give you a way in yet. Small businesses and freelancers using Adobe Creative Cloud alone cannot use the agent through ChatGPT or Copilot today — it is an enterprise-platform feature, and the price of entry is the platform itself.
Administrator approval is required to set up access, and that is not a formality. The administrator decides which users get access and what they can see, and those decisions carry through into the third-party assistant. The read-only behaviour of the assistant integrations is a comfort here: even a user with broad Adobe permissions cannot accidentally rewrite customer data by asking Copilot a careless question. But the administrator still has to do the provisioning work, and in a large organisation that can take days or weeks of internal approvals.
Cost is the awkward part. Adobe disclosed no pricing for Adobe CX Enterprise Coworker in this announcement — not a per-seat price, not a platform fee, not even a range. Usage draws on a shared AI-credit pool, but Adobe has not said what the pool costs or what happens when it runs out. If you are budgeting, the honest answer is that you cannot budget from public information. Ask Adobe (or your Adobe account team) for the credit pricing before you promise a rollout date to your own leadership.
What Adobe Did Not Announce
A launch announcement is also a list of silences, and this one has several worth naming plainly.
No pricing. As above, there is no price for the agent, no price for the integrations, and no published credit cost. For a product aimed at enterprise buyers, that is normal — enterprise software is usually quoted, not listed — but it means no reader of this article can answer the most basic buying question: what does it cost?
No adoption or usage figures. Adobe said nothing about how many organisations use Adobe CX Enterprise Coworker, how many queries it handles, or what results those organisations have seen. The product has been generally available since June 2026, so five months of usage exists somewhere — but none of it is public. Independent analysts noted the same gap: coverage of the announcement pointed out that Adobe disclosed no adoption figures and no named customer deployments for the agent.
No named customers. The announcement names nine systems integrators and six agency groups, but not a single customer brand using Adobe CX Enterprise Coworker inside ChatGPT or Copilot. The integrators — Accenture, Capgemini, Cognizant, Deloitte Digital, EY, IBM, Infosys, PwC and TCS — are building ready-to-use agentic solutions on top of the agent for specific industries. The agencies — dentsu, Havas, Omnicom, Publicis, Stagwell and WPP — are personalising it with their own intellectual property, agents, tools and industry expertise for joint clients. That ecosystem is real and impressive, but it is a promise about the future, not evidence about the present. A list of partners is not a list of users.
No governance detail. Adobe says Adobe CX Enterprise Coworker works “with human oversight”, but the announcement does not explain how actions taken from a third-party assistant are governed or audited, or whether features are identical across all five platforms. The one confirmed difference — no Firefly image generation through Claude — suggests the platforms are not identical, and buyers should ask for the full comparison before choosing where to deploy.
No dates for the three coming-soon platforms. Amazon Quick, IBM and Snowflake are “expected soon”. In enterprise software, “soon” can mean next month or next year. Do not plan a Snowflake-based rollout around it.
None of this makes the announcement unimportant. Distribution to five major assistants in one day is a genuine event. But the gap between “available” and “proven” is where smart buyers live, and right now Adobe CX Enterprise Coworker sits firmly on the “available” side of that line.
Why This Matters: Agents Go Where the Work Already Happens
The deeper story behind Adobe CX Enterprise Coworker is not about Adobe at all. It is about where AI agents live.
For the first wave of enterprise AI, the answer was: inside the vendor’s product. You bought the platform, your team learned its interface, and the agent worked there and nowhere else. That model had a cost nobody talked about much: every new agent meant a new tab, a new login, a new habit. A marketer’s day became a tour of dashboards, each with its own assistant that knew only its own data.
The second wave — the one the Adobe CX Enterprise Coworker announcement belongs to — reverses that. The agent goes to the user. If a marketer already lives in Copilot because their company runs on Microsoft 365, the Adobe agent meets them there. If an analyst lives in Claude, the agent meets them there. The assistant becomes the front door, and the specialist agents behind it supply the domain knowledge: Adobe for customer experience, Databricks for data, Snowflake for analytics.
This is also why the timing matters. The same month, other parts of the industry moved in the same direction: Walmart opened its store to shopping agents, Amazon let customers text an AI agent to shop, and some major websites started blocking agents they did not control. The two trends are opposite sides of one argument. Some companies are opening their doors to agents; others are locking them. Adobe has chosen the open side, and it is using open standards to make that choice hard to reverse: once your agent works through MCP and A2A, pulling it back behind a wall would be a visible retreat.
There is a competitive edge to the move as well. Adobe’s rivals in marketing technology — Salesforce, HubSpot, Oracle — all have their own agents, and all of them face the same question: do you keep the agent inside your platform, or do you send it out? Adobe has answered first and loudest, with five platforms at once. The risk of answering first is that you expose Adobe CX Enterprise Coworker to comparison: a marketer using Claude can now compare Adobe’s answers with whatever else Claude can do, side by side, in the same window. The reward is distribution — and distribution is how enterprise software has always been won.

Adobe CX Enterprise Coworker vs the Alternatives
If you are deciding whether Adobe CX Enterprise Coworker is the right agent for your team, the honest comparison is not against ChatGPT or Claude. Those are the rooms the agent now works in, not rivals. The real comparison is against the other enterprise agents that want the same job: orchestrating customer experience work.
| Agent / product | What it is best at | Where it lives | What to know |
|---|---|---|---|
| Adobe CX Enterprise Coworker | Customer lifecycle orchestration: campaigns, audiences, journeys | Adobe CX Enterprise, plus ChatGPT, Claude, Databricks Genie, Gemini for business, Copilot | Needs Adobe Experience Platform + Real-Time CDP, Journey Optimizer or Customer Journey Analytics; no public pricing |
| Snowflake CoCo | Enterprise data coworker: asking questions of governed data | Snowflake | Strong on governed data and SQL-backed answers; narrower marketing focus — see our Snowflake CoCo explainer |
| Zoho Analytics Agentic Analytics | Business-intelligence agent with an MCP server | Zoho Analytics, plus agents that speak MCP | Lower entry price and a write-capable MCP server; built for smaller teams — see our Zoho explainer |
| Claude Message Threads | Long-running, branching agent conversations | Claude | A general-purpose agent workspace rather than a marketing specialist — see our Claude Message Threads explainer |
The table makes the trade-offs plain. Adobe CX Enterprise Coworker is the specialist: it knows customer journeys, audiences and campaigns in a way a general assistant does not, because it sits on Adobe’s experience data. Snowflake’s CoCo is the data specialist, Zoho’s agent is the accessible one for smaller teams, and Claude’s threads are the general workspace. Most large organisations will end up using more than one of these, which is exactly why the MCP and A2A standards matter — they are what let the specialists cooperate instead of competing for the same window.
For a team already invested in Adobe Experience Platform, the decision is straightforward: the agent is an extension of a system you already pay for, and the new access removes the friction of opening another tool. For a team that is not on Adobe’s platform, the decision is much harder, because the price of entry is the platform, not the agent. In that case, the right question is not “is the agent good?” but “is the whole Adobe stack worth it for us?” — and that is a much bigger conversation.
What to Watch Next
Five things will decide whether the Adobe CX Enterprise Coworker announcement is remembered as a turning point or a press release.
1. The three coming-soon platforms. When Amazon Quick, IBM and Snowflake actually go live, and whether they launch with the same features as the first five, will show how fast Adobe can extend the pattern — and whether the “expected soon” language was optimism or a real schedule.
2. Pricing. The first public price — or the first credible leak of one — will tell smaller enterprises whether this agent is for them at all. Until then, the addressable market is effectively “organisations already on Adobe Experience Platform”.
3. Feature parity. The Firefly gap in Claude is one known difference. If more gaps appear — a connector that works in Copilot but not in Gemini, for instance — buyers will start choosing platforms by capability, and Adobe will face pressure to close the gaps.
4. Named customers with measured results. The first brand that says, publicly, “we ran this inside Copilot for a quarter and it saved our team X hours” will do more for adoption than any number of partner announcements. Watch Adobe’s own customer stories over the next two quarters.
5. Governance answers. As agents start acting across systems through the new MCP connectors, the audit question gets sharper: who approved what, and where is the record? Adobe’s “human oversight” line needs a concrete mechanism — approval steps, logs, revocable permissions — before regulated industries can deploy this broadly.
If you are evaluating the agent for your organisation, the practical next step is small: ask your Adobe administrator whether Coworker is provisioned, pick one platform your team already uses daily, and run a two-week trial on real questions — campaign triage, audience forecasts, journey conflicts — with the people who would actually use it. A trial on real work will tell you more in fourteen days than any announcement can.

Frequently Asked Questions
Is Adobe CX Enterprise Coworker free?
No public pricing has been disclosed. Adobe CX Enterprise Coworker is an enterprise feature of Adobe CX Enterprise, and usage draws on a shared AI-credit pool. Adobe has not published what the credits cost. Ask Adobe or your Adobe account team for pricing before planning a rollout.
Can I use Adobe CX Enterprise Coworker in ChatGPT without an Adobe account?
No. You need an Adobe identity with the right permissions, and your organisation must have Coworker provisioned plus Adobe Experience Platform and at least one of Real-Time CDP, Journey Optimizer or Customer Journey Analytics. The ChatGPT plugin is a doorway, not a standalone product.
Does the agent change my customer data when I ask it a question in Copilot or Claude?
According to Adobe’s setup guidance, the assistant integrations return read-only insights and do not modify customer data. The newer MCP connectors are described as able to “take actions” across systems, so confirm with Adobe which actions are live and whether they need approval before relying on that.
Why is Firefly Image Generation missing from Claude?
Adobe’s own setup documentation states that Firefly Image Generation is unavailable through the Claude integration. Adobe has not given a reason or a date for adding it. If image generation matters to your workflow, use one of the other four platforms.
When will Adobe CX Enterprise Coworker come to Snowflake, IBM and Amazon Quick?
Adobe says availability in those three is “expected soon” and has given no date. Treat them as a roadmap until Adobe announces a launch.
How is this different from asking ChatGPT or Claude a marketing question directly?
A general assistant answers from its training and the web. Adobe CX Enterprise Coworker answers from your organisation’s own Adobe customer data, brand rules and channel history, with the permissions of your Adobe identity — that context is what makes Adobe CX Enterprise Coworker different from a general chatbot. The difference is context: your data, your brand, your customers — not generic marketing advice.
The Bottom Line
Adobe CX Enterprise Coworker is now where marketers actually work. On 4 November 2026, Adobe put its customer-experience agent inside ChatGPT, Claude, Databricks Genie, Gemini for business and Microsoft Copilot, built on the open MCP and A2A standards, with new connectors reaching into Amazon Ads, Meta, Snowflake Cortex AI and more. For organisations already running Adobe Experience Platform, it removes the biggest friction in enterprise AI: opening another tool. The agent comes to the assistant your team already uses, answers from your own customer data, and — for now — reads without writing.
What would make this announcement a genuine turning point is still missing: public pricing, named customers, measured results, and a clear governance story for actions taken across systems. Until those appear, the right move is a small, honest trial on real work — not a company-wide rollout on the strength of a press release. The agent is available. Whether it earns its place in your team’s day is something only your own two-week test can answer.
Sources: Adobe newsroom, 4 November 2026 — “Adobe Integrates Agentic CX Enterprise Coworker Into AI Platforms“; Futurum Group, 4 November 2026 — “Adobe Enterprise Coworker Expands to Third-Party AI“.
