// LAB NOTES
Meet the Lab
Who we are, and a few of the things we have built.
The short version
- This is a research lab. We build instruments: tools made to test an idea and learn something true.
- We work across many areas: documents, health signals, public records, factories, and the design of AI itself.
- We have built dozens of instruments. Below are a few flagship ones.
- These notes come from this lab. We build the same way we teach here: test it, measure it, write it down.
What the lab is
We are Celaya Solutions Research (CSR), an independent lab based in El Paso, Texas.
We did not start in software. We spent years in heavy industry: running data centers and building high-voltage equipment. That work taught us one thing that shapes everything we do now. The same patterns show up everywhere. In wiring, in sound, in code, and in how a mind sorts the world. Once you see a pattern in one place, you start to see it in all of them.
So we build instruments, not products. An instrument is a tool made to learn something, not to sell something. We aim for surprise. We want to find out what is true, even when the safe move would be to ship something simple.
One method runs through all of it. Treat the work like a lab. Form a guess. Test it. Measure what happens. Write it down so anyone can check it. That is the same method these notes teach you.
A few flagship instruments
These are a handful of the tools we have built. Each one is its own project.
RECALL. Document intelligence. It reads a large pile of documents, then lets you ask plain questions and get answers that point back to the source. We proved it on a public archive of more than 22,000 Texas history documents. It ran at almost no cost, and it produced a receipt that shows the records were never changed.
EPPE, the El Paso Proof Engine. Civic proof. It takes an everyday record, like a utility bill, and stamps it onto a public ledger. Later, anyone can prove the record is real and was not changed. It is built for plain accountability, starting at home.
MORTEM. Live body data. It streams heart and body signals onto a dashboard in real time. It can also hand that data off in the same format hospitals use, so the readings are not locked inside one app.
CORTEX. Factory intelligence. A team of software agents that work together to make sense of how a plant runs. It came straight out of our years on the factory floor.
CLOS. A cognitive operating system. A larger group of agents that run as one system. It is our test bed for a hard question: how do many small parts add up to something that thinks?
BYTEFRAME. A lean AI engine. A new way to build a language model that is small enough to run on your own machine, not just in a giant data center. It is part of how we study running AI in private, on hardware you control.
These six are a sample. The lab runs dozens more. The wide range is on purpose. Working across many fields is how we find the patterns that tie them together.
Where to begin
New here? Start with the note on token cost. It covers the basics of using AI without wasting money.
After that, the notes build in order: writing better prompts, picking the right model, running your own private model, and building systems that do many steps at once.
Every note ends with steps you can run yourself. That is the promise, and it is the whole reason this lab writes anything down.