An AI Collaborator inside Adobe Workfront does not open a chat window. It logs in, picks up an assigned task, pulls the brief and brand guidelines already attached to the project, and returns finished work to a queue where a person signs off. Adobe moved the capability into production on August 13, 2026, four months after unveiling the broader agentic vision behind it at Summit in Las Vegas. Full production rollout to every Workfront customer lands October 15.

The product is simple to describe. An AI Collaborator is a permissioned user inside Workfront, the kind of account a project manager or copywriter would hold, except the work behind it runs through an agent instead of a person. Assign it a task and Workfront hands over context automatically: task description, campaign brief, brand voice, messaging rules, competitive analysis, whatever the project already has on file. The agent does not need someone to craft a careful prompt. It needs the same access to project context a new hire gets on day one.

The Feature Is Small. The Precedent Is Not.

Adobe frames AI Collaborators as a workflow fix, closing the gap between agents working invisibly in someone's chat window and the rest of the team seeing that work land. The change reaches further than that description suggests.

Workfront extended its identity and permission model to a non human actor. The agent gets scoped access, a place in the review chain, and a logged history of every step it took, the same governance apparatus that already covers the people who work inside the platform. The engineering choice was structural: extend the platform's existing definition of who, or what, counts as a user.

That is a decision most enterprises have not made for themselves yet. Deciding what an agent can see, what it can touch, and who reviews what it produces is normally a project a security or compliance team builds from the ground up over months. Workfront customers inherit that decision already built, the moment they turn the feature on.

The audit trail is the real product.

A Narrower Slice, Shipped Faster Than Promised

At Summit in April, Adobe unveiled CX Enterprise Coworker, a persistent goal-driven agent meant to decompose a business target into a full campaign execution plan across Real-Time Customer Data Platform, GenStudio, and Adobe Journey Optimizer. General availability for that architecture was described only as "coming months." I wrote at the time that the operating model required to run it, clean customer data, a documented brand, an organization ready to work alongside an agent instead of around it, was where most enterprises were not ready.

AI Collaborators is a narrower cut of that same vision, scoped to one product instead of the full stack, and it reached customers in 115 days. That is fast for an architecture this complex, and it suggests Adobe chose to ship the governance layer first, inside a system that already had permissions and workflow structure built in, rather than wait for the harder cross-product orchestration to mature. Workfront was the easiest place to prove the model works.

Customers are connecting AI Collaborators to agents built on Microsoft Copilot Studio, Claude, and Writer, or through a public API, an MCP endpoint, or an agent-to-agent connection. Adobe's own marketing team is running the feature to generate channel-specific image renditions the moment hero assets land in a project and to draft pages in Experience Manager. That internal deployment gives the launch a second data point alongside the published beta customer testimonial.

Unknowns and Uncertainties

The published evidence of production use so far includes Adobe's own internal deployment and one beta customer testimonial. How many organizations are running AI Collaborators against live customer-facing work, and at what volume, is a question the next round of customer case studies should answer. How accountability resolves when several chained AI Collaborators pass work between each other before a human reviews the final output is also open, since Adobe describes multi-agent chaining as near-term roadmap direction rather than a shipping capability. Pricing and licensing terms for AI Collaborators seats were not part of the announcement.