If you’ve ever sent a client a preview link for a new website design, you’ll know what comes back. It’s typically a screenshot with a red circle on it, an email saying the “second bit” doesn’t look right, or a request for a call about a change that could have just been an email.
By the third round of revisions, nobody’s sure which comments apply to which version.
Jumplinks Flow moves that conversation onto the page itself. Reviewers open the rendered page, highlight a sentence or click on an image, and leave a comment anchored to that exact spot. The page then moves through clear review states until someone approves it.
What Jumplinks Flow Brings to WordPress
Flow started life as an editorial workflow plugin for agencies, modelled on the way developers review code with Git. A page is drafted and sent for review, the reviewer leaves inline comments, the author makes changes, and the page is approved and published.
The team behind it, Jumplinks Systems, then realised the same merge-request style of review suited client work just as well. Developers could send a client a link, collect feedback pinned to the live page, and ship once the client signed off, with no WordPress account needed on the client’s side.
The latest addition is support for the WordPress Abilities API and MCP, which exposes Flow’s review data to AI agents. An agent that builds a page can now read the comments a human leaves on it, make the fixes, and send it back for another look. It’s the angle getting the most attention right now, and with good reason.
Where Jumplinks Flow Fits in a WordPress Stack
Flow sits alongside your website builder, with dedicated integrations for the block editor, the Classic Editor, WooCommerce, Elementor, Bricks, Beaver Builder, Divi, Avada, Breakdance and Oxygen. Reviewers never touch the builder since they only comment on the page rendered through your theme, with the real header, footer and spacing that any visitor will see.
There’s an obvious comparison here with hosted feedback tools like BugHerd, Marker.io, and Atarim. The difference with Flow in particular is that it stores comments, statuses and review history in your WordPress database, and its only external call is to Gravatar for reviewer avatars.
How Jumplinks Flow Works in Practice
To use Flow, you assign a reviewer from within WordPress, and it can be either an existing user or an external email address that receives a private magic link. The reviewer comments on the page and either approves it or requests changes, and you resolve the threads and resubmit.
Comments Stay Where They Belong
Highlighting a sentence or clicking an image, video, or embed attaches the comment to that exact piece of content. Threaded replies and resolved comments keep the revision history next to the content so you always know what each one was referring to.
Approval That Gates Publishing
With mandatory review switched on, the publish button stays locked until a reviewer approves the change. This block is enforced by Flow on the server, so it can’t be bypassed through the REST API, which becomes even more important once AI agents are involved.
A Feedback Queue Your AI Agent Can Read
An agent connected through MCP or the WordPress Agent Connector can send a page into review, then read the comment queue once a human reviewer has left feedback. Each comment arrives with the exact passage it was anchored to, so the agent isn’t left guessing which heading “tighten this” refers to.
The agent has to post a note explaining what it changed before it can resolve a comment, and it’s instructed to ask in the thread when a comment is unclear. Plus, they have no ability to approve, so the final sign-off always sits with a person for that ever-so-important human-in-the-loop.
Jumplinks has tested seven setups, including the Oxygen and Breakdance Agent Connectors, the Elementor MCP beta, AI Engine, WPVibe, Easy MCP AI, and Novamira. You can see the loop in the AI demo video below, and there’s a full breakdown on the AI Ready page.
What Stands Out for Agencies, Developers and Editorial Teams
If you run an agency, the per-site pricing is the stand-out benefit of using a tool like Flow over other competitors. Flow Pro doesn’t count reviewers, so you can invite as many clients and stakeholders as a project needs, and features like multiple reviewers with a minimum approval count and site-wide review suit the sign-off rounds most agencies typically run.
For freelance developers, the free version already covers sending a client a magic link and working through their feedback like a merge request, and it’s a far more precise way to review AI-built pages than typing corrections into a chat window.
Meanwhile, if you run an editorial team, which is where Flow’s story began, every review decision is recorded in WordPress instead of scattered across email threads and shared documents.
Getting Started With Your First Review
The quickest way to judge Flow is on a live project. Install the free plugin, pick your content types in the settings, and assign a page to your client’s email address so they get a magic link. Switching on mandatory review from day one means nothing goes live before you sign-off on it.
If you already build with an AI agent through a supported connector, you can then let it work through the client’s comments itself. When you’re ready to scale, Pro adds multiple reviewers, site-wide review, and webhooks for n8n, Zapier, and Make.
Jumplinks Flow Pricing and Licensing
The core review workflow is available in the free Flow version. Upgrading to Flow Pro gives you per-site licences with the same features on every tier:
- Single site: $49 per year
- 10 sites: $89 per year
- Unlimited sites: $199 per year
Monthly billing is also available, and there’s a 15-day money-back guarantee in case it’s not the right fit for your projects.
Is Jumplinks Flow a Good Fit for You?
If you build WordPress sites for clients and you’re tired of chasing feedback across screenshots and email, Flow is worth trying on your next project, and the free version will tell you quickly whether the workflow suits you.
For anyone already letting AI agents build pages through one of the supported builders or MCP plugins, it’s going to instantly fill a gap that’s easy to overlook, since giving an agent more capabilities doesn’t help much without a precise way to tell it what it got wrong.
