Pivot Toward AI and Agents
AI is unleashing a lot of potential. I used to just dabble in it, as it was an impressive tool for starting new projects or building PHP pages. However, Opus 4.5 and Claude Code have been eye-opening. Then, using the API key with OpenClaw was very interesting, but I quickly got into the weeds because of how expensive it is via the API. The drama and restrictions of using ClaudeMAX keys with OpenClaw and other tools are justified, since the value proposition of the subscription is too good.
Regardless of concerns, the excitement makes me want to dive deep into the fight and go super hard.
I'm currently in a weak position with my home lab setup since I had a house fire, and I just have a laptop. I've been using Claude Code to do research, build mobile games, build desktop apps, and more. It is truly an amazing technology. My current feeling is to not trust the siren call long-term and to focus the next few months on key endeavors. I have also run into headwinds in a few areas, which I intend to talk about now.
In this post and in this future blog, I intend to write about my intentions, goals, and thoughts.
Future Endeavors
In terms of making games, I have made a few, which I'll document in a later post in more depth as I figure out what I want to do with them. However, I asked Claude and leveraged Ralph to produce a report on AI making games. It confirms my biases from trying things and also leveraging tools like meshy. The report is https://nexivibe.com/research/gaming.web/index.html, and it is good. When I have Ralph use multiple personas in a loop, the results for research are rather good. I will be building an index of all the research I do and have it posted on this site as I build it up. This research was a splash of cold water.
In terms of making tools, I am finding myself biased away from the web and toward building tools that help me get stuff done using the machine. A useful tool that I made, called "cutter," lets me load an image, associate metadata with regions within the image, and then cut the image up. I'll extend this tool in a few ways, but the idea is that I can take an AI image with smaller parts and then cut them up. This also provides me a place to generate cleanup and other useful image manipulation. The other, and more useful, tool that I have made is Deckify, which lets me build decks of cards from images or image sets with tables. It's already an amazing tool that can be used to take source images (and future text) to then stamp out a lot of cards. A long-term goal I am taking on is to build software that makes it natural to make card games via The Game Crafter.
In terms of Adama, I have plans to get AI to be able to generate Adama scripts, as AI has told me that Adama is useful for AI agents. I'm not 100% sold that this is the case, but if I can teach AI to make Adama agents, then I could have it spin up 100 different games and realize the board game infrastructure play. Maybe AI lied to me about the AI agent plan, but I also see that if I don't have a way for AI to leverage Adama, then I'm screwed no matter what. I have found a measure of happiness in leveraging Adama for my own projects and also keeping costs under control. Furthermore, if I can teach AI to know Adama, then this lowers the barriers for others to leverage Adama in a meaningful way.
In terms of writing, I have a rather large corpus of writings that I am going to assemble. I had Claude rewrite my first book, which I am going to read (I already have a printed edition to fully experience). I also had Claude do some research and build a Q&A template for writing a new book. Now, AI solves my biggest problem of writing a book: I can't really enjoy reading what I created. It's the ultimate paradox. I like writing, and I have figured out how to get AI to leverage my voice (the benefit of having a large corpus of writing to draw from). As a side note, this document is my real voice with only guidance on grammar from AI using my marketing department tool. I use AI to project my voice out to social media to tune it to the audience and platform, since I don't get traction on my own voice.
This year, my ranch will be built, and I can hit the ground running in 2027 raising cattle. Using Claude and two hours of Ralph, I got research on automated ranching at https://nexivibe.com/research/ranch.web/index.html. In short, the research is amazing, and this helps me start to fully digest what I'm about to engage in. I can use this research to make capital investments to optimize my time and learn the ins and outs of ranching. Until then, I have about a year to learn AI as fast and completely as I can.
My life is ultimately at a crossroads while I figure out my priorities. The challenge now is focusing, since the potential to get distracted is absolutely massive. The manic energy that I have had in the last week is truly next-level. So, here is what I am telling myself.
The Priorities as of Today, Which May Change at Any Moment
The absolute priority is still health. I finished December strong, then got sick in January, and now I'm making sure I fully recover. I have noticed the dry air plus extreme exercise is not good, so I need to heal my throat more fully. I intend to fully leverage AI, and I will share all the things I discover here, since AI makes health a "no excuses" kind of endeavor.
First, I want to finish collecting all my writings into a unified repository such that I can fully cook my voice and style. I will also take the time to answer the 63 questions that AI generated to help me write a new book, so I want to explore this in more detail. Since this can burn a lot of tokens, this is going to be a strict weekend activity (or when I am rate-limited—who knows).
Second, I am going to dip my toes into having Claude document and start understanding Adama from the bottom up. This will inform me of how well I can succeed in having Claude be able to generate .adama scripts. I will also start vibing an MCP server and see if it lives up to my standards. I have a litany of ideas to explore, and I want to push some limits.
Third, I'm going to stop messing around with mobile apps entirely, since the games are not going to be professional and my app ideas are low-value-add to my life right now. The biggest lift I can imagine is having the Hevy app help me better with workouts, so I have done the principal engineer thing and sent them feedback. If need be, then I can spend a week to vibe code an exercise app. However, I would want an Adama backend and an RxHTML dashboard.
Fourth, I'm going to stop focusing on games for a few months beyond teaching the AI about Adama to generate board game backends. Ultimately, I have a working game of sorts, but the rules need a lot of love. If I can get Adama to make backends, then I can adapt my client runtime as well and will have an AI-ready game ecosystem where AI can generate 3D games. Using AI to generate things solves a key challenge I have with games: being surprised.
Fifth, I will continue to do deep research endeavors using Claude, as I intend to better understand the world with critical executive summaries of... everything. These will be posted here, and I'll build an index and maybe a research tool as well.
Sixth, I'm going to stop investing in OpenClaw and build out Adama as my own orchestration and agent technology. The security concerns are interesting at the moment, and I don't see how the existing integrations improve my life dramatically.
Seventh, I will avoid investing in AI hardware, as the current landscape is not profitable and the next generation of hardware will be much better. I am deciding to not go deeper into the tools side of learning and tweaking models at this time, since that requires more specialization for AI games than I want.
As a philosophical bent, I am happy to not review mobile code that gets generated. However, I will continue two endeavors that are human. First, I will write fully human words here to help keep my mind sharp. Second, I will be very strict with how Adama gets updated, with my code reviews and even writing manual code.
With the power of AI able to potentially handle the productive side of being human, it begs the question of what it means to be human. The answer to this riddle now seems obvious to me: you have complete control over your destiny. The future of humanity will be people having complete control to manifest themselves or to become entirely consumed. The human experience of learning and manifesting joy while provided for by machines seems like a new way to be. I'm not 100% in this, but it does feel like we will have new options. The social games we play will also be different, so the future is exciting.