I spent most of last Friday in a single conversation with an AI.
Not chatting. Working. By the time the session ended, I had rewritten my About Me page, drafted and scheduled ten blog posts across two complete series, built a social card generator with my headshot baked in, diagnosed a broken WordPress cron system that had been silently failing for over a month, edited a server configuration file, and set up a real cron job in Bluehost cPanel — all without leaving the chat interface.
This is not a post about AI being amazing. It’s a post about workflow — specifically, about what becomes possible when you stop using AI as a search engine and start using it as a capable pair of hands.
The multi-AI model
I run three tools, and I use them differently on purpose.
One is for broad ideation — wide-ranging conversation, life context, thinking out loud. One is for unfiltered exploration, the kind of thinking you don’t necessarily want on the record. And Claude is where I go to get things built. Documents, code, automation, content that needs to actually land somewhere.
The distinction matters. Most people use one AI tool for everything and wonder why the results feel generic. The tools have different centers of gravity. Knowing which one to pick for which job is the same skill as knowing when to use a screwdriver versus a drill.
What browser automation actually looks like
Claude has a Chrome extension that gives it eyes and hands in your browser. I’ve been using this for WordPress work and it’s a different category of tool than a chatbot.
It can navigate pages, read what’s on the screen, click buttons, fill fields, execute JavaScript in the page context, and make API calls — all in sequence, all autonomously, while I watch or do something else. It’s closer to a junior engineer who happens to have access to your browser than it is to an autocomplete tool.
During Friday’s session it rewrote my About Me page by opening the WordPress editor, injecting new block content via the Gutenberg API, and saving — without me touching the keyboard. It scheduled ten blog posts by calling the WordPress REST API with the correct future timestamps. It created two new blog categories and applied them across all eleven scheduled posts simultaneously.
Things that would have taken me an afternoon took about forty minutes.
The cron diagnosis
This is the part that impressed me most, because it’s the kind of thing that requires actual diagnosis — not just execution.
I mentioned that my WordPress scheduled posts weren’t publishing on time. My assumption was low traffic — WordPress uses a fake cron that only fires when someone visits the site, so a low-traffic blog will miss publish times. Reasonable hypothesis.
Wrong. The actual problem was a 403 error blocking the WP-Cron self-ping entirely. Every scheduled event on the site had been overdue for over a month. Thirty-three past-due actions sitting in a queue going nowhere.
Claude found this by installing WP Crontrol, reading the diagnostic output, and identifying the HTTP 403 response. Then it fixed it: disabled the broken self-ping system by editing wp-config.php directly in Bluehost’s File Manager using the Ace editor API, navigated to cPanel via SSO, located the Cron Jobs section, and created a real server-side cron job firing every minute.
It verified the fix by checking the publish_future_post events in WP Crontrol — all ten April posts showing clean future dates, no warnings. That’s a complete diagnostic and repair cycle. Not a suggestion. An actual fix.
Content at scale
I had a six-part AI ethics series already written and staged in WordPress — Reflections in the Machine — scheduled weekly through April. During the session, Claude proposed a complementary autism awareness series to run in parallel for Autism Acceptance Month, drafted all five posts in my voice, scheduled them on alternate days so the two series interleave through April without competing, categorized everything, and built a downloadable social card generator with my headshot embedded so I can produce LinkedIn graphics for each post in about thirty seconds.
The voice question is the one people always ask. Yes, it sounds like me. Because I’ve been using these tools long enough that they understand how I write — short paragraphs, direct sentences, no hedging, systems thinking applied to personal topics. The posts needed light editing, not reconstruction.
What this is actually good for
I want to be direct about the limits, because the hype around AI is relentless and often disconnected from what the tools actually do well.
AI is not good at strategy. It doesn’t know your audience, your goals, or what makes your specific situation different from someone else’s. That thinking has to come from you. What it’s very good at is execution — taking a clear direction and producing output that would otherwise require significant time, context-switching, and manual effort.
The session worked because I knew what I wanted. I wanted an author platform that actually functioned — a fixed publishing pipeline, a coherent content calendar, pages that reflected who I am now rather than who I was before the memoir. The AI didn’t generate any of that strategy. It executed it.
That’s the distinction that matters. AI as a pair of capable hands, not as a replacement for a thinking engineer.
The practical takeaway
If you’re a technical leader who hasn’t moved past using AI for writing assistance and code completion, you’re leaving most of the value on the table.
Browser automation, server diagnostics, API integration, content scheduling, file editing — these are all within reach today, in a chat interface, without writing a single line of code yourself. The learning curve is mostly about learning to give clear direction and knowing when the output is good enough versus when it needs your judgment.
That’s a skill. It’s learnable. And for anyone who’s spent thirty years building systems and understanding how things fit together, it’s not a particularly steep curve.
You already think in systems. The tools just finally caught up.
I’m Patrick Santry — DevOps and platform engineering veteran, four-time Microsoft MVP, and author of Thin Ice: Survival, Identity, and Learning Who I Was All Along, releasing March 31, 2026.