Log Entry #4
Time Spent Testing: 40 Minutes
Managing multiple WordPress sites can often feel like a long list of repetitive, manual tasks—logging in, updating content, and navigating between dashboards. Naturally, I started wondering: What if there was a way to simplify this entire process with AI?
That’s where Claude AI and the WP MCP Server came in. I set out to test a proof of concept: could I manage bulk updates across multiple WordPress sites using a spreadsheet and a bit of natural language magic? The answer? Yes—and it was surprisingly fun. Here’s how it all unfolded.
The Experiment: Setting the Stage for WP Bulk Updates
For this test, I created a small WP multisite with three WordPress sites: lisablog, bartblog, and helloblog. Each site was connected to Claude through the WP MCP Server enabling structured communication via the WordPress REST API. Need a guide on setting up WP MCP Server? Check out my WP MCP Server blog post.
To explore different workflows, I tested two approaches:
- Google Sheet Workflow: Exporting a spreadsheet to structure bulk updates.
- Natural Language Prompts: Bypassing spreadsheets entirely and using conversational commands to make updates.
Both methods worked, but I found myself drawn to the simplicity and efficiency of natural language updates. There’s something deeply satisfying about crafting a prompt with plain descriptions and what you want and watching it happen—no spreadsheets, no dashboards, no extra clicks.
The Process: Step-by-Step
Step 1: Checking Access to Sites
After adding the new websites credentials to my json file I prompted Claude to confirmed access to all three sites—lisablog, bartblog, and helloblog—and identified a shared endpoint for updates: /wp/v2/pages.
Step 2: Using a Google Sheet for Bulk Updates
For the first test, I created a Google Sheet with columns for site names, page URLs, and content updates. It looked something like this:
| Site | Page URL | Content Update |
|---|---|---|
| lisablog | /sample-page/ | Replace squirrel poem with cat poem |
| bartblog | /hello-goodbye | Create new page with stray cat poem |
| helloblog | /sample-page/ | Remove squirrel poem |
Claude processed the spreadsheet and executed updates accordingly. Each action returned a status code 200, confirming success.
Step 3: Testing Natural Language Updates
Next, I skipped the spreadsheet and tried a conversational approach. Here’s an example prompt:
"Update the following WordPress sites:
lisablog: Replace squirrel poem on /sample-page/ with a miniature pony poem.
bartblog: Add a new page called 'memecat' featuring a meme cat poem and an image.
helloblog: Add an interactive button game to /sample-page/.Claude executed the requests seamlessly:
What I Learned when using WP MCP Server with Claude AI:
This experiment wasn’t just a technical success—it highlighted the potential of AI to assist in website management in ways that are both practical and creative.
1. Natural Language = “intuitive” alternative
Writing commands in plain language and watching them executed flawlessly removes so much friction. It’s the kind of UX improvement that makes you wonder why we haven’t always done it this way.
2. Spreadsheets are Still Useful
For tasks requiring consistency and scale, spreadsheets are invaluable. They provide structure and control, especially for complex updates.
3. AI as a Collaborative Partner
Claude didn’t just execute commands—it guided the process, suggested structures, and identified the right endpoints. This dynamic made even the most technical steps feel accessible and manageable.
4. Creative Potential is Endless
By streamlining repetitive tasks,this experiment unlocked creative opportunities—like adding interactive elements or designing entirely new pages. It’s a reminder that AI isn’t just about efficiency; it’s a tool for creativity, too.
AI WordPress Multisite Management
This proof of concept is just the beginning. Tools like Claude and the WP MCP Server have the potential to revolutionize WordPress management by making it more efficient, intuitive, and enjoyable. Imagine being able to:
- Scaling Up: How does this approach handle dozens—or hundreds—of sites?
- Automation: Can these workflows integrate with tools like project management platforms or automated triggers?
- Memory and Context: How far can natural language commands go, and how can we use AI memory effectively without overloading it?
The possibilities are endless— and there’s still work to do, of course. For example, how do we balance simplicity with complexity as workflows scale? And how do we ensure that AI remains a tool for empowerment, not dependency? These are questions I’ll keep exploring in future experiments.
