5 Memory Types

UAML organizes knowledge into five distinct memory types, each serving a different purpose. Together, they give your AI agent a rich, structured understanding of the world.

📖 Episodic Memory

Records of specific events and experiences. "What happened in yesterday's meeting?" Timestamped, contextual, searchable by time range.

📚 Semantic Memory

General knowledge and facts. "Python 3.13 removed the GIL." Organized by topic, searchable by content, confidence-scored.

⚙️ Procedural Memory

How-to knowledge and workflows. "To deploy, first run tests, then build, then push." Step-by-step procedures your agent can follow.

🔍 Reasoning Memory

Decision traces and logic chains. "I chose option A because of X, Y, Z." Provenance you can audit and verify.

🔗 Associative Memory

Links between related knowledge. "This bug is related to that config change." Graph-based connections that surface relevant context automatically.

How It Works

Each memory type maps to the 5-layer data architecture. The memory type determines how knowledge is stored, indexed, searched, and shared.

from uaml.facade import UAML

uaml = UAML()

# Episodic — store an event
uaml.learn("Team decided to use Rust for the parser",
           topic="decisions", source_type="meeting")

# Semantic — store a fact
uaml.learn("UAML uses SQLite for local-first storage",
           topic="architecture")

# Search across all types
results = uaml.search("storage decisions")
            

Why It Matters

Better recall — the right memory type improves search relevance
Context-aware — episodic memories carry temporal context
Auditable — reasoning memories create decision trails
Connectable — associative memories surface hidden relationships

← Back to UAML