Chronicles
From the MyRSF.net archives of the Email Chronicles
Ask AI Anything About The RSF Association
Author: Phil Trubey
Date: February 24, 2026
Hello, Phil Trubey here, sending an occasional email about RSF.
HOA Assistant
For a while now, I've been wanting to do a programming project just so I could try out the new Artificial Intelligence (AI) programming tools.
So I decided to write an "ask me anything" chatbot that had been trained on the RSF Association's governing documents, Board resolutions and Davis-Stirling HOA law ensuring the bot knows all about the Protective Covenant, our Regulatory Code, the golf and tennis club operation plans, etc.
You can try it out at my website, https://myrsf.net, just select HOA Assistant from the menu or home page.
You can ask it questions, and even follow-on questions, meaning you can have a conversation with it. Just finish all your questions with the return/enter key.
There's a bunch of operational stuff it doesn't know, like personnel names, contact info, forms, etc. All that stuff is best asked of the Association directly.
But you can ask it questions like:
How can I build a fence?When does the Association send out the budget?How are dues calculated? Can I use stone as an exterior cladding material for my house?
I think it's pretty good, but if you notice any weird answers, please let me know so that I can see where the confusion came from and correct it.
How It Works
All the documents (about 16 PDFs) are parsed by a custom program and headings are extracted. I used an AI to do this parsing and it figured out that the Regulatory Code has chapters and convoluted paragraph headers, while the Protective Covenant just has paragraph numbers, and the Bylaws are organized in yet another way.
This structured data is then fed into a special database used for AI called a vector database. The structured English text is "indexed" by concept inside this database.
When you ask a question, the question is indexed just like the text of those 16 documents and similarity-by-concept matches are found in the vector database. This results in about a dozen matches across all 16 documents related to your question. These matches and your question are then all fed into a powerful AI Large Language Model which then writes out an answer.
The actual AI processing to answer your question is done at a large data center somewhere and takes about 2 to 8 seconds on a shared supercomputer. Currently I'm using OpenAI's GPT 4.1 model which isn't even state of the art, but works well anyway.
How "I" Wrote It
What I really wanted to do with this project was use the latest Claude Code AI tool to help me write all this code (about 1,200 lines of Python with some front end HTML/CSS).
And ironically, in the end, I didn't write a single line of code.
I started off talking with Claude Code (CC) describing what I wanted to do. It gave me three options, and I picked the "hardest" one (meaning the one that needed the most code to write and would give me the most flexibility to change things).
And then CC just kept doing things for me. It would say, "OK, the next step is I look at your PDFs and figure out their structure", and I would say, "Go ahead", and then it would say, "OK, I've found all the headings, want me to write a program to extract all the headings and text?", "Sure, sounds good to me!". And so on.
When I gave it PDFs that had been scanned and thus were just images, no embedded text, it asked if it could install an Optical Character Recognition program. It installed it, noticed the installation errors (there's always installation errors, or missing dependencies), fixed the errors, used the program and kept on chugging.
Something that would have taken me weeks to do manually (since I would have to learn a bunch of new stuff), took me a day of on and off "work" on my part.
Computer programmers are the first profession to embrace this AI technology, making the average professional programmer easily ten times as productive when writing any kind of computer code.
Where Is This Heading?
While computer programming is the first industry to get disrupted, it certainly isn't going to be the last.
Theoretical physicists are using the latest powerful foundation models to help them with their work. You know, the people trying to figure out what Einstein couldn't (unification of General Relativity and Quantum Mechanics).
The only reason AI hasn't completely taken over the rest of white collar work yet is that there simply hasn't been enough time to train it on any one organization's particular work flow or internal documents. Also, the foundational models are improving so quickly that nothing's been stable long enough for larger organizations to adopt anything lest they end up with an obsolete solution in 6 months.
But it's obvious that AI disruption is going to wash over our entire economy like a tsunami; it's just a matter of where and when.
It won't happen overnight or in 2 years. History has proven that highly disruptive technologies (electricity, the personal computer, the Internet, smartphones) have a 10-20 year adoption cycle before they seep into most of the economy.
I think AI diffusion will be similar, there are just too many human induced friction points in our economy for it to move faster, no matter how useful it is.
Brave New World
OK, so what does the world look like 20 years from now?
First, we'll be much richer (see my X post argument about that: https://x.com/PTrubey/status/2026367710314246223?s=20).
Second, I predict leisure industries will grow. Just as AI-based education is giving students a lot of extra time to do other activities, AI-based business will be so efficient, a 2 day work week is in the cards. We're slouching our way there anyway through "work" at home programs.
Third, what is going to replace people's desire for purpose and achievement that is currently supplied through work? I suspect people are going to realize there is one thing that AI/Robots are never going to be able to do, and that's having kids.
Parents consistently say the most fulfilling thing they've ever done is raising kids. I suspect we're going to see a re-emergence of parenthood as the next "fad" that'll sweep through our social landscape.
Predicting the future is always fraught with danger and these predictions should be taken with a grain of salt. But AI's impact on our society will almost certainly be as profound as any change in the history of civilization.
Previous emails, Rancho information, poll results, and the old RSF Post archives available at myrsf.net. If you aren't on my email list and want to receive my occasional scribblings, go to this web page and click on the "click here to subscribe" link after the preamble. To unsubscribe, click the link below.