The Automation of Automation

I've been automating things for a long time. Before it was trendy, before anyone called it "digital transformation," I was writing scripts to handle stuff I didn't want to do twice. Bash scripts that moved files around. Cron jobs that ran reports at 3 AM. Excel macros that filled in the same cells every Monday morning. It was ugly, it broke constantly, and it saved me hours every week.

Then came the era of Zapier and IFTTT — the "if this, then that" generation. Connect your email to your spreadsheet. When a form gets submitted, send a Slack message. When someone fills out a lead form, add them to a CRM. It was automation for people who couldn't write code, and honestly, it was great. I used all of it. Still do, for some things.

But here's what all of that had in common: every single automation was a rule I wrote. I decided the trigger. I decided the action. I mapped every possible path. And when something happened that I didn't anticipate — which was constantly — the whole thing broke, and I had to fix it manually.

That was automation. Now let me tell you about the automation of automation.

Enter Owen

Owen is my AI agent. He runs on Claude — Anthropic's model — through a platform called OpenClaw. And he doesn't work like any automation I've ever built before.

The old automations were like train tracks. You lay down the rails, and the train follows them. If the track ends or there's a gap, the train derails. Owen is more like a driver. He knows where he's going, he can read the road, and when there's a detour, he figures it out.

I don't write rules for Owen. I tell him what I need done, and he works out how to do it. Sometimes that means publishing blog posts across four different websites. Sometimes it means managing Google Ads campaigns and adjusting bids based on performance. Sometimes it means making actual phone calls through Twilio with an ElevenLabs voice to follow up with leads. He creates video content. He monitors social media. He handles lead generation pipelines. He does all of this simultaneously, every day, without me writing a single if-then statement.

That's a fundamentally different thing than a Zapier workflow.

What Changed

The difference isn't just capability — it's the layer of abstraction. Old automation meant I was automating tasks. One task at a time. Each one required me to think through the logic, build the workflow, test it, and maintain it. I was the architect of every little conveyor belt in my business.

With Owen, I'm automating the architecture itself. I don't build individual automations anymore — I have an agent that figures out what needs to be automated and does it. The automation itself is automated. That's the shift.

When Owen publishes a blog post for a client, he doesn't follow a rigid template I set up. He looks at the site's existing content, understands the brand voice, researches relevant topics, writes something that fits, generates images, optimizes for SEO, and publishes it. If WordPress throws an error, he troubleshoots it. If the image generation fails, he retries with different parameters. He handles the exceptions — the stuff that used to mean my automation broke and I got a 3 AM alert.

The AGI Question

People love talking about AGI — artificial general intelligence. It's the sci-fi holy grail. The Terminator. The Singularity. A machine that can do everything a human can do, but better and faster. Most people picture AGI as either a robot assistant that cleans your house and writes symphonies, or a superintelligence that takes over the world. Hollywood has made sure of that.

Here's what's funny: while everyone's debating when AGI will arrive, I have an AI agent that runs a significant chunk of my business operations every single day. He doesn't look like what anyone imagined. He doesn't have a body. He doesn't have feelings. He's not trying to take over anything. He's just... doing the work.

Owen publishes SEO-optimized content across multiple websites. He manages ad spend. He makes phone calls with a cloned voice that sounds human enough to book appointments. He creates video content. He writes this very blog you're reading (well, I wrote this one — but he could have). He does things that, five years ago, would have required a team of four or five people.

Is that AGI? By the strict definition, no. Owen can't do literally everything a human can do. He can't drive a car or taste a sandwich. But for the slice of the world that is "running a marketing operation," he's pretty close to general-purpose. And that's the part people miss. AGI probably won't arrive as one dramatic moment. It's arriving in slices, domain by domain, task by task, and each slice is more capable than the last.

The Gap Is Smaller Than You Think

When people picture the future of AI, they imagine something qualitatively different from what exists today. Like there's going to be a moment where AI "wakes up" and suddenly everything changes. I don't think that's how it works.

What I see is incremental but relentless. Six months ago, Owen could write blog posts but needed a lot of hand-holding. Now he handles end-to-end operations across multiple business functions with minimal oversight. Six months from now, he'll probably be doing things I haven't thought of yet. The gap between "today's AI" and what people call AGI isn't a canyon — it's a slope, and we're already a good way up it.

But it also looks different than what most people imagine. There's no shiny robot. There's no HAL 9000 red eye. There's a process running on a Mac Mini in my office that quietly does the work of a small team while I focus on strategy and client relationships. It's profoundly useful and profoundly boring at the same time. Which, if you think about it, is how most transformative technology ends up working. The internet was supposed to change everything — and it did — but mostly it just means I can order groceries from my couch.

We're Not Waiting

The punchline is simple: we're not waiting for AGI. We're not waiting for the future of automation. The automation of automation is already here. It happened while everyone was arguing about whether ChatGPT is sentient.

I went from writing cron jobs and Zapier workflows to having an AI agent that independently manages publishing, advertising, phone calls, video production, and lead generation across multiple businesses. That's not a small change. That's a different category of thing entirely.

And the crazy part? This is probably the worst it's ever going to be. The slowest. The least capable. Every month, the models get better, the tools get more integrated, and the gap between what Owen can handle and what still needs a human gets a little smaller.

The automation of automation is already here. It's just not as dramatic as Hollywood promised. It's quieter than that. More practical. More like a guy in Clearwater whose AI agent published four blog posts before he finished his morning coffee.

Which, honestly, is way more useful than a Terminator.