Paradox Systems
Machines that are
taught, not scraped.
We build artificial intelligence from the ground up, then teach it the way a person is taught: subject by subject, by someone who knows the material, tested after every lesson. Every product in this company runs on that idea.
Advancement
The work is the model, not the wrapper.
Most software calling itself AI is a thin layer around somebody else's model. The intelligence belongs to the vendor, the bill arrives every month, and the moment the API changes the product changes with it.
We took the slower road. The model is ours, the tokenizer is ours, the corpus is ours and the teaching pipeline is ours. That means we can say precisely what our system was taught and what it was not, and no third party can change its behaviour with a version bump.
Integration
One intelligence, every product.
Advancement without integration is a research project. Every application here is a place where the same taught intelligence does real work: quoting a job, organizing a drive, reading a load, watching a device, valuing a codebase.
The apps are not separate experiments that happen to share a brand. They share the framework, the model and the teaching method, which is why an improvement in one shows up in all of them.
The intelligence
PAIMON
Model 1 of machine teaching. A system that knows what it was taught because somebody sat down and taught it.
PAIMON is our own language model, trained from scratch on a curriculum rather than scraped from the open internet. It learns one subject at a time, is tested after each lesson, and answers from the material it actually holds. When it does not know something, that is a gap in the curriculum, and gaps can be fixed.
It is the intelligence underneath the rest of this page. The applications in the catalogue are where it earns its keep.
The catalogue
What we build
Twenty-six products across six groups, plus the framework and the model they run on.
The framework
3AIS
Advanced Automated Augmented Intelligence System. The structure that turns a model into a system.
A model on its own answers questions. A framework decides what it is allowed to perceive, how it reasons about what it found, which actions it may take and who approves them. 3AIS supplies those pieces: perception, cognition, ethics and execution, with a Cross-Core Engine correlating the data underneath.
It is the difference between an assistant that talks and a system that runs something.
Together
3AIS and PAIMON, integrated
The framework handles structure. The model handles language and judgment. Put them together and you get a system that can read a city's open data feeds, notice that two of them disagree, explain the disagreement in plain language and route the decision to a person who is allowed to make it.
That combination is what a city deployment actually is: 3AIS deciding what to look at and what to do, PAIMON explaining it to the humans in the room. The same pairing runs underneath the internal operating layer and the smart city packages.