I’ve recently started an open-source experimental project:
https://github.com/Aminmasoud123/Reactive-AIThe idea is to explore an AI architecture where the system continuously observes changing signals and maintains evolving internal states, rather than treating every input as an isolated event.
The basic concept is:
Changing Signals
↓
Internal State
↓
Measure Change
↓
Update State
↓
Reaction
↓
Updated State
↺
The current prototype is based on simple mathematical signals, but I’m interested in exploring whether the same idea could be applied to real-time crypto data, such as price, volume, volatility, or order-book changes. I’m curious whether this approach could be useful for things like anomaly detection, market signals, or real-time risk analysis. I know there are related approaches such as state-space and recurrent models, so I’m mainly experimenting to see if this particular idea leads anywhere interesting.
Any feedback or suggestions are welcome
