Had.

Every week, there was a new "game-changing" AI tool that was going to solve the problem of the day.

Jasper for copy. Hootsuite AI for social. Superhuman for email efficiency.

The reality? A good process, clean data, and ChatGPT gets me 80% there.

Here's the painful detail on $7,337.47 worth of tools I no longer use, so you can learn from my mistakes.

My 2024 AI tool spending hit over $7K in subscriptions I ultimately ditched. Turns out I'm not alone, 53% of software licenses go unused, wasting massive budgets on tools nobody actually uses.

Shock Therapy: I uploaded my Amex statements to ChatGPT and had it act as a financial analyst to categorize my tool costs. The results were embarrassing. 🫠

Jasper, Copy.ai, Rytr – Expensive Mad Libs: I tested all these tools early, but after the initial wow factor, their generated copy felt formulaic and soulless, like filling in blanks on a pricey Mad Libs.

I forgot to cancel some of them… 🤯

Why I went back to ChatGPT + Custom Instructions: Ultimately, I dropped the copy-specific tools and standardized on ChatGPT (Plus version) with carefully written prompts. It turns out a well-crafted prompt in ChatGPT can produce content just as good as Jasper or Copy.ai – without yet another $49–$99/month subscription.

I then use Claude to edit or refine an initial outline or research from ChatGPT.

Hootsuite’s, Blaze.ai, TweetHunter - $2,208 of Automation without strategy: These promised to auto-generate and schedule “engaging” posts for every channel. In reality, auto-posting isn’t a strategy – it’s just activity. Then I noticed my metrics and engagement was dropping.

I was shipping slop.

My audience could tell something was off. And they’re not alone – people can easily tell when content is AI-generated and often react negatively. I actually saw lower engagement on posts that were mostly AI-created versus ones I crafted. The algorithms didn’t boost the machine-made posts either; some social media platforms even seem to penalize “auto” content.

🤔 Lesson Learned: A robot impersonating a human on Twitter without a plan isn’t winning any fans.

I reverted to a hybrid approach. Now I use AI to brainstorm ideas or research topics (ChatGPT with search is great for this), but a human (me or a team member) reviews, adds context, and schedules the posts thoughtfully.

💡 Pro-Tip: We also post natively to avoid any algorithmic down-ranking.

I hate managing email; six accounts daily means six streams of spam and cold outreach. I tried everything, but realized I was solving the wrong problem.

Superhuman - $480/year. For optimal performance, you need to use their UI. I've saved $40/month using ChatGPT, Zapier, and Gmail for four accounts. (The other two accounts have IT departments that won't let me connect AI.)

Leonardo - $288.

I wanted to test Midjourney competition. This isn’t it.

And with ChatGPT's improving image capabilities, even Midjourney might get cut soon…

I now upload PDF’s from QuickBooks and Amex to ChatGPT each month to keep track of my tool spending. If I’m testing a tool for you all, I do it for a month or two and then cancel it. I also set up dedicated labels in Gmail to keep track of each tool so they don’t get lost in the inbox maelstrom. 

Reply with your biggest AI tool disappointment. Let's learn from each other's mistakes.

P.S. Need AI help? book a 2hr 1:1 with me to get your AI upskilling roadmap sorted.