Adobe's Agent Ambitions: From Creative Tools to Enterprise AI Operating System
This is pretty interesting. Adobe, the company that’s been making Photoshop and Illustrator for forty years, now says it wants to build an “enterprise AI agent operating system.”
I read the April 20 announcement twice to make sure I understood correctly—Adobe is indeed pushing something called CX Enterprise Coworker, and they’ve assembled quite a roster of “teammates”: AWS, Anthropic, Google Cloud, IBM, Microsoft, NVIDIA, OpenAI. The lineup is as impressive as the Avengers.
From “Photo Editing Software” to “AI Operating System”
Adobe’s transformation actually has precedents.
When Firefly (generative AI tool) launched, everyone thought Adobe just wanted a piece of the AI image generation pie. But now it looks like that was just step one. CX Enterprise Coworker is positioned to help enterprises embed AI agents into the complete business process from “customer discovery” to “purchase completion.”
Translated to human language: Before, Adobe sold tools (you use them to create graphics). Now Adobe wants to sell “AI employees” (they help your customers complete the entire purchase journey).
The Payment Loop is the Key
One detail in the announcement is easy to overlook: Adobe reached deep cooperation with Adyen, PayPal, and Stripe to embed payment functionality directly into AI-driven interactions.
What does this mean?
Previously, an AI customer service agent could only “chat.” After chatting, you’d still need to jump to another page to pay. Now Adobe wants to achieve: AI understands your needs → recommends products → processes payment → completes order, all within one conversation.
This is technically quite difficult. Payments involve risk control, compliance, privacy, and AI needs to accurately understand user intent while processing transactions—the cost of errors is high. Adobe’s willingness to do this shows they have confidence.
Why So Many “Teammates”?
Looking at that partnership list, there’s an obvious pattern:
- Model Layer: OpenAI, Anthropic, Google—provide underlying AI capabilities
- Cloud Infrastructure: AWS, Google Cloud, Azure—provide compute and deployment
- Hardware Acceleration: NVIDIA—training and inference performance
- Enterprise Services: IBM, Microsoft—enterprise customer channels and integration experience
Adobe is smart. They’re not trying to handle everything themselves but positioning themselves as the “orchestration layer”—stringing together the best capabilities from each vendor and packaging them as solutions enterprises can use directly.
This strategy is similar to Salesforce’s Einstein or ServiceNow’s AI layer. Everyone is fighting for the position of “enterprise AI entry point.”
Implications for Developers
If you’re doing AI application development, Adobe’s move has several points worth noting:
- Vertical integration is more valuable than general capabilities: Adobe’s advantage lies in its forty years of creative content accumulation, knowing what enterprise customers really need
- Ecosystem partnership is mandatory: The era of going solo is over. Your product must seamlessly integrate with existing enterprise IT architecture
- End-to-end experience is the key differentiator: Just doing chat isn’t enough—you need to extend AI capabilities to business closure
My Personal Take
Adobe’s ambitions are big, but the challenges are significant too.
The enterprise market isn’t like the consumer market—decision cycles are long, customization needs are high, and stability requirements are extremely demanding. Although Adobe has brand advantages, in the “enterprise software” field, they’re still a new player.
That said, seeing an old-school software company willing to go all-in on AI rather than resting on its laurels is admirable. After all, there are too many “PPT transformation” cases in the industry—Adobe is at least actually doing things.
Whether it succeeds or not, we’ll have to wait and see.