Fine Print

Updated November 16, 2026

Where the work has to be

A brand check can check only the work that it can reach. Most products check work inside their own application, so someone must put the work there first. Some products go to the work: they run in the editor of the writer, or an agent sends the work to them.

What are the places where a product checks work?

The products in the directory check work in six places:

  • Its own library. The work is an asset in the product. Bynder checks assets in its DAM (digital asset management) system, and Cloudinary checks assets in its Media Library.
  • Its own editor or template. People make the work inside the product. Marq checks projects in Marq, and Adobe GenStudio scores the ads that it makes.
  • A review in its application. Someone adds a file to a review, and reviewers open it there. Filestage, Ziflow, and PageProof work this way.
  • The tool where a person writes. An add-in or a browser extension checks the text in place. Markup AI works in Google Docs and Word, and Lytho has extensions for Chrome and Edge.
  • Any place that an agent or your code can reach. The agent that made the work sends it to the check. Rainbrand, Markup AI, IntelligenceBank, and Lytho document this route.
  • The live web. The product scans pages after publication. Siteimprove and PerformLine work this way.

Why does the place matter for work that agents make?

An agent makes many drafts in a chat or a code editor, and most drafts never reach a library or a review. If the check is in a library, a draft gets no check until a person uploads it. If the agent can request the check, it can revise the work before a person sees it.

The route for the agent matters too. Some products let an agent start the check on work that the agent sends. Other products let an agent start a check only on work that is already in the product, or only read a saved result.

Where does each product check?

A highlight marks a limit that the vendor states.

Product Where the work must be Can an agent run it?
Adobe Content in AEM as a Cloud Service Yes, through the AEM MCP server
Artwork Flow Files in Artwork Flow Not published
Bynder Assets in the Bynder DAM Partly: its API runs it on Bynder assets
Cloudinary Assets in Cloudinary Partly: its API reads saved results
Frontify A chat in Frontify guidelines No: its MCP server only reads guidelines
IntelligenceBank Its platform, add-ins, or extension Yes, through its MCP server or API
Lytho Lytho proofs, browser, Canva, Office Yes, through its Claude connector
Markup AI Google Docs, Word, or any agent Yes, through MCP or its API
Rainbrand Where a connected agent makes it Yes, through MCP or its API
Writer Writer and its extensions No: its MCP server runs playbooks
Filestage Files in a Filestage review Partly: its API adds a file to a review
PageProof Proofs in PageProof, and design plugins No: its MCP server reads metadata only
Ziflow Proofs in Ziflow No: its API has no ReviewAI call
Marq Projects in Marq No: its MCP tools have no brand check
Templafy Word, PowerPoint, Excel, and Outlook No: its MCP server makes documents
PerformLine Live web, social, email, and calls Yes, through its API
Siteimprove Websites that Siteimprove crawls Partly: its API runs policy checks

What do you ask a vendor?

  • Where must the work be before the check can run?
  • Can an agent or our own code start the check, or can it only read a result?
  • Does that route need a higher plan or an add-on?