I Move Fast. That Doesn't Mean I Stop Thinking Critically When Using AI

When I was little, I was nicknamed "Speedy". It was the first indicator of my later ADHD diagnosis in that I couldn't sit still, but also that I was on the move, I wanted to get places, even if it meant that I just crawled as fast, and recklessly as possible to get there. It's also why I procrastinate on things I feel like I cannot complete in an expeditious fashion. I worry I will get discouraged, frustrated, or miss a step in a long drawn out process. Quite ironic for someone who thrives in a professional environment of problem diagnosis, mitigation and root cause review. It just goes to show how our personal weaknesses drive our professional strengths.

Needless to say, I've postponed really working on my website for more than a few months. I've known that my website wasn't doing what it needed to do at this point in the business development journey. Not in a vague, "I should update this someday" way — in a concrete, how-is-anyone-supposed-to-hire-me way. There were no real service descriptions, just the placeholders I posted there at launch. Someone could land on this site, read every page, and still not be entirely sure what I actually do or how to work with me. And I was simultaneously trying to get serious about SEO, which meant that whatever I built next, I needed to build it right the first time.

The podcast was part of the same problem. It was linked on a general page on the website but every episode lived somewhere different — Spotify, YouTube, scattered links with no central home. If someone wanted to find an old episode or share one, they were on their own. For something I plan to keep building, that wasn't sustainable.

So I sat down this past week with a few different AI platforms, settling on Claude for assistance with the website build and got to work. Not because I suddenly had extra hours in the day. Certainly not because I was overcontrolled and refusing to pay someone to do it for me, there is no budget for that right now even if I wanted to. But because the problems were real, the window was there, and I'm working on not waiting for perfect conditions.

What thinking critically with AI actually looks like

I want to be specific here because I think there's a lot of noise around what AI is and isn't good for as you heard on the recent Motivation N'at episode with Jim Jordan, and I've developed a pretty clear opinion through actually using it. Keep in mind, this is my perspective and experience, everyone should consider their unique circumstances.

I move fast - Speedy may not run anymore, but my brain has not slowed down, nor these typing fingers. That's not a flaw I'm managing — it's how I operate. But moving fast has always had one real cost: the time it takes to track down step one, step two, step three just so I can get to step four, which is the part I actually need. Reading through platform documentation, hunting down the right support thread, figuring out which version of the instructions applies to my setup — that's where I used to lose time. I'd have little to show for the amount of work it felt like I had put in.

AI compresses that. I can describe exactly where I am and what I'm trying to do, and get pointed directly at the relevant piece of the manual. That's not replacing knowledge. That's getting to the work faster so I can apply the knowledge I already have. What is critical is that I am honest with myself if I do actually need to go back to step one and that I ask for that "hold up and back up" if needed.

What I don't use it for — and what I'd caution anyone against — is turning off your own judgment. Because the moment you do, you miss things. For example, Claude suggested some really ugly colors to use as accent colors and despite those colors passing accessibility requirements, they weren't something I was willing to accept.

Don't scrap the page

Here's a concrete situation depicting what I mean.

While building out the podcast item section, I ran into a problem with my CMS dynamic connectors. They weren't pulling the right episode to the right page — every link was routing to the same episode regardless of which one you clicked. I went back and forth with Claude, providing screen captures and architecture and pathways. In the end, nothing was working. Claude told me to go "nuclear" and pull the page down and start over. Two and a half hours of work lost. Not to mention, I still wasn't clear on what I botched in the first place, and there was a chance a rebuild to end up the same way. Was it user error, or did Claude direct me wrong simply because perhaps it thought I was on a different version of wix.com's editor.

So, I sat with that for a second. And decided I'll look at one more option, and if that didn't work, I'd accept that I'd have to build manual links and slowly work on a rebuilt over time.

I pulled everything up next to each other and just looked at the facts in front of me. I had a live URL for each of the episode pages. I had a place for that live URL to connect to. If I put that URL in manually into a button not related to a dynamic connector the link would work. So the only piece missing was that the link was auto updating on each page like I needed.

So I looked at what the CMS actually allowed. I knew I could add manual URL entry fields, that's how I was linking YouTube, Spotify, etc after all. What would happen if I simply put the autogenerate URL into a new field where it looked like a static line item entry? It should work just like the other links. I built a button connected to that field instead of the broken dynamic one, and it worked.

Not the elegant solution. But the right one for where I was. And I only found it because I didn't accept "start over" as the only path forward.

Trust the tool. Verify the output. Keep your own eyes open. AI doesn't know all of the workarounds because it's not there with you running the tests. It can help you eliminate options, or prioritize which options to test, but you know best what you are aiming for.

This isn't a forever plan — and I'm not pretending it is

I'm not delusional: doing this myself right now is a function of where I am, not a declaration of how I'll always operate. There will be a season where budget allows for someone who does this full time and does it better. I genuinely look forward to that season, my other half probably does as well given my mutterings and late nights working on this.

But "not ideal conditions" is most of the time, for most of us running businesses on our own terms. The choice isn't between perfect and nothing. It's between moving forward thoughtfully or not moving at all.

AI is one of the tools that makes thoughtful forward motion possible when resources are tight. It's not a shortcut around expertise — mine or anyone else's. It's a way to apply expertise more efficiently, to read the manual in a fraction of the time, to compare documents and identify tangible differences, to provide first level insights. The user must then take on the responsibility of identifying which portion of the manual they need most urgently, to validate if the differences are real, or just surface level, to take the first level insights and layer in that varied expertise and knowledge. There's a difference, and it matters.

What I'd tell anyone sitting where I was

Work on your prompts. Be specific about your situation, your platform, your constraints. The more context you provide, the more useful the output. A vague prompt gets you a generic answer — and generic answers are where people get into trouble, either blindly following something that doesn't fit or dismissing the tool entirely because it missed the mark. Ask the tool how confident it is that it can execute the task you've provided. Encourage it to ask further questions prior to providing a response, nine times out of ten, you've forgotten a key component of the task.

Push back when something doesn't feel right. Ask it to look at the problem a different way. And when you find a better solution on your own — which you will — that's not a failure of the tool. That's you doing your job. I then ensure the tool knows there was an option it didn't consider - this is how we contribute to better tools.

Moving fast and thinking critically aren't opposites. I'd argue you can't really do one without the other.

Now, go check out the rest of the website, peruse the first set of episodes that have been backloaded into the repository, and share this post, my services or the podcast with someone that might find it useful.