@tjrobertson52

If you're like me, you're seeing a lot of people do really cool things with AI recently and you can start to feel like you're being left behind. But I wanna talk about why I think 99% of us actually don't want to be on the bleeding edge of AI right now.

Now, to be clear, if you do any kind of knowledge work and you're not using AI on a daily basis right now, I do think you're about to be left behind. What I'm talking about are these people automating complex workflows, building their own models, and more recently using claudbot to do some really cool things.

I think I've gotten pretty good at using AI. I think it makes me two to 10 times as productive as I used to be, depending on the task. But looking at what some of these people are doing, it's clear they're moving a lot faster than me.

But here's the thing, we're seeing a biased sample. We're not seeing the person that spent 100 hours putting together that complex automation just to have it break or never work in the first place or get replaced by some simple software that came out two months later.

Now with Clibout, we are hearing a lot more horror stories, but these risks have always existed and the primary risk of trying to stay on the bleeding edge of this technology, just wasting a bunch of your time. That being said, AI continues to move very fast, it can already save us a ton of time and I do think it's important that we keep up.

So let me talk about the heuristic I use to determine if something is worth my time. I think the main question we're trying to answer is how long would it take to build the system and how much time is it gonna save me in the long run? To figure that out, we need to understand a few things like what are the models actually capable of and what are they likely to be capable of in the future?

If you try to automate something with AI that it can't yet do reliably, well, you're gonna waste a lot of your time banging your head against the wall and then ultimately you're still just gonna end up hiring someone to do the work instead. On the other hand, there have already been hundreds if not thousands of startups that have spent millions of dollars creating software only to see them become irrelevant overnight when OpenAI, Google or anthropic add that capability to their model.

So personally, whenever I think of something really cool I'd love to build with AI, I ask myself, is this something that would just be obviously beneficial to millions of people? If so, there's a pretty safe bet that there are already 100 startups building the exact same thing or it'll probably just end up being a core feature of one of the major models. So instead of pouring weeks into building the perfect version of it, I can just spend an afternoon putting together some prompts, templates, and SOP's for my team.