This year, I’ve been building an AI social media production pipeline to promote the 25+ shows on the Autistic Culture Podcast Network. It took me a week of effort, but I managed to program Claude Code to talk to NotebookLM, which now feeds our YouTube, Instagram, and TikTok pipeline, and a task that used to take me a full day now runs in the background. The first time I ran a real batch through it, I felt the guilt of always being behind fade away. I could physically feel my nervous system recalibrating. Maybe for the first time since I was 10 years old, with these artificial intelligence agents, I could recapture the feeling of not being behind!
And then, the other shoe dropped. The next week, the automatic processing didn’t work. I had to rebuild the agent from scratch. Meanwhile, I was in the middle of building my next agent to send out press releases when new shows launch, but that got put on hold. At about 3 a.m., I was in a full-on AI coding haze when I ran out of credits and had no choice but to go to bed. At that moment, I realized all this time I was trying to save by using AI was sure taking up a lot of time.
Business culture has one instrument for measuring AI, and it’s the stopwatch. Productivity!
Minutes saved. Output multiplied. Projects completed. Revenue up.
Every dashboard asks how much faster the company is moving. Are we working less, though? Because I, for one, have been working more and with a panic about not being left behind. Getting more done is not the same as making the work day less demanding.
The gap matters even more for neurodivergent founders like me. A report published in June by the U.K.’s Lilac Centre, based on a national survey of more than 600 neurodivergent entrepreneurs, found that 76% started their businesses specifically to work in a way that suited them, but 79% report workload-management or burnout challenges as a result.
The same report calls AI and digital tools “accessibility scaffolding,” and I’d back that up. I even wrote a book way back in 2023 about how AI could be used as an accessibility tool for neurodivergent people. I use AI daily to start tasks I’d otherwise stall on, to turn scattered thinking into a structured draft, to get past the executive-function wall that shows up when my monotropic flow is not flowing in the directions I need it to. I do not question that AI can take tasks off your plate. But, what I missed in my book, and I think others are missing, is the cost of using AI.
Call it the Automation Tax. It’s all the extra work you have to do to make the AI make your job go faster, and the expectation that now you can do things faster, you still need to use every working hour to get even more done. The speed got turned up, but the workload didn’t decrease.
Building the Autistic Culture Podcast Network pipeline taught me the size of that tax. Wiring three systems together, learning where each tool tends to fail so I know what to double-check; none of that showed up in an “hours saved per post” calculation.
Meanwhile, my workload never seemed to decrease. The hour saved was filled with the next task, because faster output means more time to plan and promote. Instead of writing my magnum opus or learning Italian with the time I saved, I found myself managing more agents.
This is why when I use a tool now, I ask three questions. What exact task disappears? What new work shows up in its place? And do I really end up with fewer things to manage, or just different things?
A single AI-drafted Instagram caption might save me only five minutes, but it removes the dread of staring at a blank box until the moment to post has passed; for a task that small, that’s an excellent return. Meanwhile, generating five newsletters in the time it used to take to write one still leaves five newsletters to fact-check, schedule, and defend if something in them is wrong. Output goes up, sure, but so does the workload.
If I can’t answer that last question: Do I really end up with fewer things to manage or different things, then the tool has just shifted or even increased the workload.
The best AI tool isn’t the one that helps you produce the most. It’s the one that frees you up to do more of what you want to do.