Back to technical architecture

SEMANTIC UNDERSTANDING

Semantic understanding

Resolve what entities, terms, and constraints mean for a specific team at a specific time, beyond keyword matching.

Interpret the context, not just the words

The same phrase can mean different things across teams, time periods, and tasks. “Activity fell recently” could refer to this week or to a monthly trend. “Active” might mean logging in, completing a core action, or making a purchase. An agent needs to resolve the business concepts, effective dates, and task constraints before choosing its evidence.

Represent business state in records people can check

Enterprise semantic memory can bridge gaps between databases, business semantic layers, and retrieval systems. It does not replace those systems. It organizes the changing context an agent needs to make a decision: the full metric name, its source, where it applies, when it took effect, and who approved the change.

  • Separate concept from state: distinguish what a metric is from which version applies at a given time.
  • Keep the boundaries: treat team, product, time period, and access as conditions on use.
  • Surface uncertainty: when two definitions could be valid, ask instead of choosing the closest textual match.

A useful test is whether this reduces metric-definition errors and makes results easier to verify.