AI agent learning

Enhance AI agent learning through episodic memory, enabling agents to improve by learning from past successful interactions.

Episodic memory overview

The represents historical experiences and their associated learnings in the form of feedback. It's the ability to reflect on and learn from them by applying memory of similar episodes to new experiences. This capability enables agents to extract meaningful patterns and insights from experiences and use them effectively in the future.

Understanding episodic memory in AI agents

When an agent learning is enabled, the AI agent:

• Focuses on learning from execution outcomes, building episodic memory of what went wrong and how it was fixed.
• Acts as a feedback refinement loop, converting agent failures into structured learning for future tasks.
• Memory is heavily based on experience correction rather than just context extension.
• Memory entries carry relevance weights (like attention scores) → aligns with adaptive retention seen in human-like learning.

To configure agent learning at the AI agent level, see Define the specialty of an AI agent.

To configure agent learning at the AI Agent Studio level, see Set up long-term memory.