LLMCAMP
← All resources Masterclass replay Agent memory

Build AI Agents That Actually Remember

Understand agent memory, then build a support copilot with OAMP, the OpenAI Agents SDK, and Oracle AI Database.

Hosted byRichmond Alake
1 hr 7 min

The full session

Ideas → Demo → Code
Watch at your own pace. Build along with the notebook. Watch on YouTube

Take the learning with you

Get the presentation.

Revisit the memory concepts and implementation steps in the masterclass presentation.

Open in Google Slides Public presentation · Opens in a new tab

The big picture

Give your agent something to remember.

This masterclass connects the concepts of memory engineering with a working support copilot. Richmond Alake introduces short-term and long-term memory, the four Rs, and the roles of embeddings and rerankers.

The coding session uses the Oracle Agent Memory Package (OAMP), the OpenAI Agents SDK, and Oracle AI Database to explore extraction, retention, context cards, and retrieval.

Worth remembering

Key takeaways

  1. Distinguish memory types.

    Understand short-term and long-term memory before choosing how an agent should retain information.

  2. Use the four Rs.

    Retain, recall, reuse, and refine provide a framework for thinking about the memory lifecycle.

  3. Configure what gets remembered.

    Explore extraction instructions, retention logic, and the organisation of memory by users and threads.

  4. Connect memory to the agent.

    See how tools and context cards bring stored information into a support-agent workflow.

Theory, meet practice

Inside the session

The session moves from the mental models to a working implementation, with live demos along the way.

Learn the concepts

Memory types and memory engineering principles.

Assemble the components

OAMP, models, extraction, and retention.

Run the copilot

Context cards, agent tools, and memory retrieval.

Make it work on your machine

Get the code. Build the patterns.

Follow the support-copilot implementation in the companion notebook.

Companion notebook · PythonOAMP support assistantoamp_support_assistant_example.ipynb

Follow the notebook’s setup instructions for dependencies, credentials, and database configuration.

Keep learning

Make memory your next deep dive.

Explore 100 Days of Agent Memory for more patterns, practical frameworks, and lessons from building in the open.

Explore the series
← Back to the resource library