What if every client site took one approval, not thirty logins?

Think about the last time a vulnerability alert landed for a plugin you knew was installed somewhere in your client list. Before you could fix anything, you had to work out which sites were exposed, which usually means a lot of logging in and checking one dashboard after another.

AI agents like Claude and ChatGPT can already do a lot of that work, although most WordPress MCP setups only connect them to one site at a time.

This week we came across Royally, the new cloud half of Royal MCP Pro. Your agent connects to Royally once and can then work across every site you manage, your own and your clients’, with nothing changing until you approve it.

It caught our attention for 3 main reasons:

• One request, every site. You can ask it to update meta descriptions across three sites or change a brand colour in Elementor on twelve, and it previews the change everywhere before anything goes live.
• One-click undo across the lot. Every change comes with a receipt, and if something looks wrong you can roll it back on every site at once.
• It governs your other MCP plugins too. Elementor’s MCP, Easy MCP AI and AI Engine can all run through the same connection, and every change lands in one tamper-evident audit log.

It also checks uptime every five minutes, alerts you to PHP errors, vulnerabilities and new admin accounts, and sends branded reports to your clients.

Royally launched publicly yesterday, on November 1st, with plans starting at just $9/month for 3 sites, going up to to $79/month for 100. Royal MCP Pro is included in every plan.

Check out Royally today →

We all know by now that letting an agent loose on one site is a reasonable experiment, but letting it work across thirty needs approvals and an audit trail, and that’s what Royally is built for.

So hit reply and tell me: What would you need to see before you’d let AI make changes on your clients’ sites?

Stay curious and keep building,

Mark from WP Mayor