Enhance README with memory approach details

Added detailed explanations for two memory approaches

Author: laisbsc

pai-memory/README.md

@@ -1,6 +1,20 @@
-# Pydantic AI memory examples
+# Pydantic AI Memory Examples
 
-Run postgres in docker with
+Two demos showing different approaches to implementing persistent memory for Pydantic AI agents:
+
+## Message History Approach
+### File `with_messages.py`
+Stores complete conversation history as JSON in PostgreSQL. The agent receives all past messages as context automatically, making it remember previous interactions naturally.
+
+## Tool-Based Memory Approach
+### File `with_tools.py`
+Provides the agent with `record_memory` and `retrieve_memories` tools. The agent actively decides what information to store and when to retrieve it from the database.
+
+**Key difference:** Message history provides full context automatically, while tool-based memory gives the agent selective control over what to remember and recall.
+
+## Setup
+
+Run postgres in docker:
 
 ```bash
 docker run -e POSTGRES_HOST_AUTH_METHOD=trust --rm -it --name pg -p 5432:5432 -d postgres