Episodic Memory

Facts alone don't capture experience. Episodic memory lets agents learn from past trajectories, not just static knowledge.

Episodic memory is a diary of what happened and how — 'last time the server crashed, we restarted Redis first.'

Scenario

When asked to fix a flaky test, the agent retrieves an episode from last month where the same timeout issue was solved by increasing the jest timeout.

Solution

In Agent Memory, apply Episodic Memory to this scenario: When asked to fix a flaky test, the agent retrieves an episode from last month where the same timeout issue was solved by increasing the jest timeout. Identify the inputs, run the technique, validate the output, and note one thing you would monitor in production.

Do this before moving to the next module — reading alone is not enough.

Open the Code Walkthrough below and run it locally. Change one parameter related to Episodic Memory (e.g. model, temperature, top_k, or tool name), observe the difference in output, and write 2–3 sentences explaining what changed.

Highlighted lines show where Episodic Memory happens in the code.

1# Episodic Memory — minimal example2from openai import OpenAI3
4client = OpenAI()  # create API client5
6# Ask the model to explain this topic7response = client.chat.completions.create(  # core API call for Episodic Memory8    model="gpt-4o-mini",9    messages=[10        {"role": "system", "content": "You explain episodic memory clearly."},11        {"role": "user", "content": f"What is episodic memory?"},12    ],13    temperature=0,14)15print(response.choices[0].message.content)  # show output for debugging

• Treating Episodic Memory as a black box without evaluation
• Ignoring cost and latency in production
• Skipping error handling for episodic memory

Quick recap — the most important points from this module.