Skip to main content
While a Decision represents “what must be done,” a Memory is the advisory, explanatory context that answers “why something was decided”. Memories prevent organizational memory loss by preserving the debates, evaluations, and business requirements that led to a specific architectural choice.

​
Key Characteristics

Advisory, Not Enforced

Memories provide context but are not strictly enforced by Hopsule’s AI or IDE integrations. They guide understanding rather than restricting code.

Mutable & Living

Unlike decisions, memories are fully mutable. You can update them as your project’s context or your understanding of a problem evolves.

Relational

Memories are explicitly linked to Decisions. When a developer encounters an enforced rule, they can instantly read the linked memory to understand its origin.

Semantic Search Ready

Every memory is chunked and embedded into your Project Brain using pgvector, making it instantly retrievable by the Hopper AI.

​
Decisions vs. Memories

Understanding the strict boundary between these two concepts is the key to Hopsule’s authority model.
AspectDecisionMemory
AuthorityAuthoritativeAdvisory
EnforcementEnforced by AI & IDENot enforced
ImmutabilityImmutable when ACCEPTEDMutable (can be updated)
PurposeConstrain behaviorProvide context

​
Using Memories with AI

When you package your project into a Capsule, both your Decisions and Memories are bundled together. When you ask Hopper, “Why did we choose OAuth 2.0?”, the AI doesn’t just read the rule; it reads the linked Memory detailing your evaluation of Basic Auth, API Keys, and OAuth, and provides a perfectly grounded answer.