Continual learning
How ABRIA turns user teaching into persistent semantic structure and replayable learning signals.
Source map
- python/abria_server/continual_learning.py
- python/abria_server/conversation_routing.py
Experience-first learning
continual_learning.py compiles user-authored lessons into structured examples and semantic relations. It supports broad declarative facts and property-style teaching rather than only fixed prompt templates.
Account locality
Persistent user learning is attached to the authenticated account. The ABR runtime does not automatically propagate one user's lessons into another user's cognitive fabric.
Consolidation
The service exposes continual-learning status and an explicit consolidate endpoint so learned signals can be compacted/reinforced deliberately.
Teaching vs asking
conversation_routing distinguishes user-learning turns such as 'I am teaching you...' from ordinary 'teach me X' questions, reducing accidental ingestion.