Uploaded October 2026 | Updated November 2026, 3 days ago
AI agents can read information from project notes, but that information is often lost when the conversation ends. Developers need a simple way to turn useful notes into durable, searchable memory that an agent can retrieve in future interactions.
In this session, we’ll build a practical workflow that connects a notes application to Oracle AI Agent Memory using Model Context Protocol (MCP). We’ll show how project notes can be transformed into typed memories, stored in Oracle AI Database, and retrieved when they are relevant to a later agent request.
What we’ll cover:
- Understanding the difference between conversation context and durable agent memory
- Connecting an external notes source to Oracle AI Agent Memory using MCP
- Extracting and classifying useful information as facts, preferences, or guidelines
- Storing and searching memories with the oracleagentmemory Python package
- Using a simple frontend to inspect the memory workflow
- Retrieving relevant memories across separate agent sessions
Resources:
- Slides: bit.ly/4e92XtQ
- Oracle AI Developer Hub GitHub: github.com/oracle-devrel/oracle-ai-developer-hub
- Oracle Agent Memory Package: oracle.com/database/ai-agent-memory
- Oracle MCP GitHub: github.com/oracle/mcp
- Join the Codex + Oracle AI Database Hackathon at Oracle AI World October 28th reg.rf.oracle.com/flow/oracle/oaiw26/DatabaseHackathon/page/hack
- Deeplearning.ai course: deeplearning.ai/courses/building-adaptive-ai-agents
0:00 Introduction to Agent Memory
2:36 Defining the Developer Problem
5:03 Agent Memory vs. Chat History
5:55 The Role of MCP
7:26 Oracle AI Agent Memory
8:25 Deep Data Security
9:49 Graph-Aware Retrieval
11:18 Demo Architecture and Workflow
15:30 Live Demo: Processing Memory
26:35 Visualizing Memory with Graphs
30:21 Testing Recall and Role-Based Security
34:47 Summary and Announcements
AI agents can read information from project notes, but that information is often lost when the conversation ends. Developers need a simple way to turn useful notes into durable, searchable memory that an agent can retrieve in future interactions.
In this session, we’ll build a practical workflow that connects a notes application to Oracle AI Agent Memory using Model Context Protocol (MCP). We’ll show how project notes can be transformed into typed memories, stored in Oracle AI Database, and retrieved when they are relevant to a later agent request.
What we’ll cover:
- Understanding the difference between conversation context and durable agent memory
- Connecting an external notes source to Oracle AI Agent Memory using MCP
- Extracting and classifying useful information as facts, preferences, or guidelines
- Storing and searching memories with the oracleagentmemory Python package
- Using a simple frontend to inspect the memory workflow
- Retrieving relevant memories across separate agent sessions
Resources:
- Slides: bit.ly/4e92XtQ
- Oracle AI Developer Hub GitHub: github.com/oracle-devrel/oracle-ai-developer-hub
- Oracle Agent Memory Package: oracle.com/database/ai-agent-memory
- Oracle MCP GitHub: github.com/oracle/mcp
- Join the Codex + Oracle AI Database Hackathon at Oracle AI World October 28th reg.rf.oracle.com/flow/oracle/oaiw26/DatabaseHackathon/page/hack
- Deeplearning.ai course: deeplearning.ai/courses/building-adaptive-ai-agents
0:00 Introduction to Agent Memory
2:36 Defining the Developer Problem
5:03 Agent Memory vs. Chat History
5:55 The Role of MCP
7:26 Oracle AI Agent Memory
8:25 Deep Data Security
9:49 Graph-Aware Retrieval
11:18 Demo Architecture and Workflow
15:30 Live Demo: Processing Memory
26:35 Visualizing Memory with Graphs
30:21 Testing Recall and Role-Based Security
34:47 Summary and Announcements










