Adobe launches AI collaborators in Workfront workflows

Adobe has made AI Collaborators generally available in Adobe Workfront, placing AI agents inside Workfront workflows for marketing teams.

The launch lets teams assign work to an AI Collaborator much as they would to a colleague. In Workfront, the collaborator acts as a permissioned user, drawing context from projects and returning results to the same task for review and approval.

Adobe is targeting marketing departments that already use AI tools across teams but struggle to track how that work is done. In Workfront, outputs remain visible within the existing work management system rather than being generated in separate chat tools or individual applications.

Workfront automatically packages the information an agent needs before a task is sent out, including a task description, campaign brief, messaging rules, brand voice and competitive analysis.

The feature includes role-based permissions, an audit trail and human sign-off before work is released. That structure is intended to help organisations use AI tools at greater scale while keeping records of each step.

Marketing uses

Adobe's initial use cases focus on common marketing tasks, including on-brand copy creation, localisation across markets, production of multiple ad variations and brand compliance review.

In one example, a team can connect an AI Collaborator to a copywriting agent built in Microsoft Copilot Studio so a landing page draft is created using the campaign brief and brand guidance already held in the project. That allows writers to start with a draft instead of preparing detailed prompts from scratch.

Another scenario covers localisation, where approved content can be adapted for different markets and then checked by regional reviewers before sign-off. The full path from source material to local versions remains logged in one place.

The system can also connect to a Writer agent to turn a campaign brief into several LinkedIn ad versions. For compliance tasks, a reviewer collaborator can check content before it reaches approvers and work with Adobe Brand Intelligence to reflect how a brand has been approved or rejected in past work.

Agent connections

Customers can connect AI Collaborator to agents built on platforms including Microsoft Copilot Studio, Claude managed agents and Writer. Other agents can be linked through a public API endpoint, an MCP endpoint or an agent-to-agent connection.

Adobe said setup takes only a few minutes, with users naming the collaborator, describing its role and linking it to an existing agent before assigning a task. The process is designed to lower the barrier for teams that want to test AI within established approval processes rather than through separate pilot projects.

A beta customer quoted by Adobe described the experience simply: "Feels approachable. You're making AI digestible, and usable for us."

Broader workflow

The release is part of a wider Adobe effort to tie Workfront more closely to the AI assistants staff already use. Workfront MCP lets teams access Workfront from tools such as Claude, Microsoft Copilot, Gemini and ChatGPT without leaving those assistants.

Adobe is addressing workflow in both directions. AI Collaborators send work from Workfront to agents and return the output to Workfront, while Workfront MCP lets users access Workfront from external AI tools.

Adobe's own global marketing team is building AI Collaborators to generate channel-specific image renditions when hero assets are added to a project. The tools can also draft pages in Adobe Experience Manager and handle hand-offs tied to event promotion across teams.

The product reflects a broader shift in enterprise software as vendors try to move AI features beyond individual prompts and into repeatable business processes. For Adobe, the focus is on using Workfront as the control layer for assigning tasks, maintaining records and keeping human approval in place before output is published.