docs: add AI agent memory tutorial

[owainlewis]

Summary

• Adds a new public tutorial at tutorials/ai-agent-memory.
• Writes a recording-ready LESSON.md that explains agent memory as saved context loaded at the right time.
• Adds a small shared OpenAI agent abstraction in code/agent.py:

agent = Agent(memory=StaticFileMemory())
reply = agent.run("How should I run Python commands here?")

• Adds a Python memory ladder using the same Agent shape:
  • 01_no_memory.py for current conversation only.
  • 02_static_memory.py for AGENTS.md and MEMORY.md startup memory.
  • 03_session_search.py for SQLite-backed session search.
  • 04_mem0_memory.py for external managed memory with Mem0 Cloud.
• Adds a code README, .env.example, reset helper, and focused unit tests.

Why

This brings back the AI agent memory tutorial as a clean youtube-tutorials lesson instead of a loose archived draft.

The goal is to give developers a simple mental model and runnable examples they can work through on camera:

load useful context -> call the model -> save what matters

The Agent class keeps the demo easy to explain: the agent owns the OpenAI call and short-term conversation history, while memory implementations own startup(), search(), and save().

The lesson keeps the focus on practical memory choices: static files, current conversation history, session search, app state, and managed memory. It avoids starting with vector databases or framework terminology before the core loop is clear.

Test plan

• python3 -m unittest discover -s tests
• python3 -m compileall .
• Fake-model smoke test for 02_static_memory.py
• Fake-model smoke test for 03_session_search.py
• git diff --cached --check
• git diff --check -- tutorials/ai-agent-memory
• just check blocked by unrelated untracked tutorials/herdr-agent-workflow/presentation.html in the local working tree.

Risks

• The Mem0 demo depends on the current Mem0 Cloud Python SDK shape and should be reviewed before recording.
• The examples are intentionally small teaching samples, not a production memory architecture.
• The lesson is long enough to support a full video, so it may need editorial tightening before recording.

Related issue

None.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f6e693152b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Restore the Airtable MCP tutorial files

This README was renamed from tutorials/mcp-airtable/README.md, and the same commit deletes the remaining tutorials/mcp-airtable lesson/resource files; I checked the target tree and it has no tutorials/mcp-airtable entries at all. Since this change is framed as adding the AI agent memory tutorial, existing Airtable MCP tutorial links and content disappear unintentionally; please add the new tutorial as a copy/scaffold instead of moving the old tutorial out of the repo.

Useful? React with 👍 / 👎.