Procedural Memory

Repeating the same multi-step process from scratch wastes tokens and introduces variance. Procedural memory captures proven playbooks.

Procedural memory is muscle memory — you don't think about each finger movement when typing; you just type.

Scenario

The agent has a stored procedure for 'deploy to staging': run tests → build Docker image → push to ECR → update k8s manifest → verify health endpoint.

Solution

In Agent Memory, apply Procedural Memory to this scenario: The agent has a stored procedure for 'deploy to staging': run tests → build Docker image → push to ECR → update k8s manifest → verify health endpoint. 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 Procedural 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 Procedural Memory happens in the code.

1# Procedural 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 Procedural Memory8    model="gpt-4o-mini",9    messages=[10        {"role": "system", "content": "You explain procedural memory clearly."},11        {"role": "user", "content": f"What is procedural memory?"},12    ],13    temperature=0,14)15print(response.choices[0].message.content)  # show output for debugging

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

Quick recap — the most important points from this module.