Tools, cost, and what’s next

Oh, night and day. I have said many times that agentic coding and AI in general makes your impossible list possible. That couldn’t be more true for me. If you look at my GitHub activity, every year I’ve had an account is almost nothing, and then this year it explodes. That’s entirely because of agents. Being able to build solutions that far exceed my own skills has been kind of intoxicating, frankly. Almost too much, like a rabbit hole you dive down and fall way, way into. The speed and ability to create these things has been a huge unlock. None of this would exist otherwise. It would all be on my list of impossible projects.

In money, with AI, I have subscriptions to multiple services, ChatGPT and Claude, and those add up. I also have costs in AWS to host all the projects I’m building and running. In time, there’s a continuous amount I put into publishing the Weekly Thing, and then project time on top of that.

But both the cost and the time connect back to a core thesis I have about learning. I believe strongly that to have a career in technology is to have a career in learning. When I’m building, I’m learning. It’s an investment in my craft and in the art of creating, and that impacts so many other things around me. I can tell you that I’m better at being a CTO because I have direct hands-on experience creating things, and I know the trade-offs and the art of the possible in a way you can’t if all you do is read and watch others. It lets me connect with my team and understand what transformation actually feels like, not just what its business results are. And frankly, it’s just what I’m passionate about. I love doing it. This feels much more like art to me than a hustle. It’s not a way to create a business. It’s a way to create something of beauty, something I see in my head that I want to instantiate in the real world.

Honestly, I’m kind of doing that right now. Road Sign Math, created in the last few days, was one of the last things on my list. I’ve worked through most of that list, and now I’ve been doing a V2 on some of those things. What I did recently with Linking Things, making it this 3D spatial kind of flyover, is one way of looking at that. I’m 100% sure new ideas will come. Just recently I had the idea that Thingy should actually be a 3D model and be able to move around and engage with people. So how do you do that? Like I said, a world that doesn’t need software is a world without ideas. I’ve got plenty of ideas.

It made the threads and eras more visible. I knew I had a MediaWiki era, or a wiki era in general. I knew I had a crypto era. And I knew the blog was the backbone of all of it, and it made that a lot more obvious. There were a handful of projects like VeloTweets that I had totally forgotten about, and it was delightful to see that come back to life.

And I absolutely adore that Things I’ve built has a place to capture artifacts of the building process, even ones that may not be part of the final website. My blog is an artifact I can read in the future. I think Things I’ve built will be a way of capturing and remembering those things in a much deeper way.

For sure, my blog. For sure, the Weekly Thing. I hope by then Another Thing has a vibrant and growing published history of new podcasts. The Thingy agent, for sure, getting increasingly more intelligent and capable to interface with everything I’ve created over the years. I’m sure 20 years from now, Elixir will be done and retired. It’s unlikely I’d be playing that game in 20 years. I think the museum will still be there. And I think I’ll have more archival sites. The Proof of a Person archive and the nnxyz museum are a new class of thing, an activity I did, with a capstone I created that will just exist in perpetuity. Almost a time capsule. I expect there will be more of those, because it’s a new lever, a new kind of thing I can do with AI that I was never able to do before. And I’m sure there will be new stuff that hasn’t even come into my mind yet.