Agent Memory

Agents without memory are stateless — every conversation starts from zero. Memory gives agents continuity, personalisation, and the ability to learn from past interactions.

Types of Memory

Different memory types serve different purposes in agentic systems.

Short-Term Memory

Conversation context within a single session. The agent remembers what was said earlier in the chat and builds on it. Lost when the session ends.

Long-Term Memory

Persistent storage that survives across sessions. User preferences, past decisions, accumulated knowledge — recalled when relevant to future interactions.

Episodic Memory

Specific past events and interactions. "Last time you asked about X, we decided Y." Gives agents the ability to reference shared history.

Semantic Memory

General knowledge and learned facts. Not tied to specific interactions but built up over time. "This user works in healthcare and prefers concise reports."

Value Pathways

How agent memory creates increasing returns over time.

Personalisation

Agents that remember user preferences, communication style, and past requests deliver increasingly tailored responses — without users repeating themselves.

Efficiency

Agents skip re-discovery when they remember context. Onboarding, troubleshooting, and routine requests get faster as the agent accumulates relevant history.

Pattern Recognition

With enough episodic memory, agents start recognising recurring issues, seasonal patterns, and systemic problems — surfacing insights humans might miss.

Institutional Continuity

When team members leave, agent memory preserves context. Decisions, rationale, and historical knowledge persist in the system instead of walking out the door.

Want agents that learn and remember?

I build memory systems across all deployment models — from cloud-native to fully isolated.