Kinsta APM and Server Logs: Tracking AI Load on WordPress

Since the arrival of the Abilities API and the AI Client, WordPress is no longer just a CMS. It has become an execution engine that connects your applications with the AI models of your choice. Kinsta shows how to monitor AI workloads in WordPress using Kinsta APM and server access logs in MyKinsta. Both are available to all Kinsta clients at no extra cost.

This shift adds a new layer of complexity, so observability and performance monitoring matter more than ever for running a modern website. Kinsta's goal is clear: give you concrete proof of how much AI features affect your site's resources, rather than just reminding you that monitoring is important.

Where AI load comes from

Your site can interact with AI in several ways. These include scans by AI bots and crawlers, internal plugin operations through the AI Client, and requests from external applications and AI agents that query your site through the Abilities API.

Take image generation or audio transcription. AI operations like these often take several seconds, and during that time a PHP worker/thread can't serve other requests, whether they come from a human user or a machine. Not every site will feel the impact. But for high-traffic business sites and e-commerce stores with hundreds or thousands of products, it's a factor you need to watch.

Two MyKinsta tools for monitoring AI

The first tool is Kinsta APM, a free Application Performance Monitoring tool in the MyKinsta dashboard. To enable it, go to Sites > sitename > APM, click Enable, and pick your preferred interval. Once monitoring finishes, you get an automated email with a link to the results page. There you'll find detailed charts for transactions, plugins, themes, hooks, slow database queries, and external requests. If you're not on Kinsta, a tool like New Relic can serve as an alternative.

The second tool is access and error logs, which are often a goldmine of insight into what's happening under the hood. These logs help you trace errors, find performance bottlenecks, and identify your main traffic sources. On Kinsta, you can download log files via SFTP or SSH, or through the MyKinsta File manager. The MyKinsta Log viewer also lets you inspect error, cache, and access logs directly.

The test: generating a featured image with AI

For the test, Kinsta asked AI to generate a featured image for a WordPress post using the official AI plugin from WordPress.org. The steps:

• Configure a connector in the Connectors screen of the WordPress admin
• Install and activate the official WordPress.org AI plugin
• Enable Kinsta APM and monitor the access logs in MyKinsta
• Create a new post, then generate a featured image using the AI feature

You'll need an API key from your AI provider, which you enter under Settings > Connectors after clicking Activate for your chosen provider. The AI plugin adds several abilities you can inspect in Tools > Abilities Explorer, including Image Generation and Editing. In MyKinsta, APM is set to a 2-hour monitoring duration, while access.log is monitored under Sites > sitename > Logs. Kinsta notes that the measurement examples come from AI operations run at different times, so recorded times may vary slightly between tests.

If you run a WordPress site with AI features, especially an online store with a large catalog, these two tools are a good starting point for seeing how much server resource AI operations actually consume.