ChatGPT Memory's scope
OpenAI's built-in memory layer for ChatGPT. Stores user preferences, context, agreed-upon working assumptions. Per-account; non-portable; OpenAI owns the storage and the format. Good for consumers wanting a persistent ChatGPT experience.
The portability gap
ChatGPT Memory doesn't leak to Claude. Doesn't leak to Gemini. Doesn't leak to your in-house systems. The memories trapped inside ChatGPT lock you into OpenAI — same pattern as Claude Projects, Gemini Workspace Memory.
GenZAgents memory snapshots
Memory snapshots are JSON exports from your AI agent's state, signed with your keypair, stored in the receipt feed. They're portable across providers (the snapshot becomes part of the portable manifest), auditable (each snapshot is a receipt), and yours (not OpenAI's).
For consumer ChatGPT use
Use ChatGPT Memory for the in-ChatGPT consumer UX (it's built in, free, hassle-free). Use GenZAgents on top for backup / portability / audit. Combine the two: ChatGPT Memory handles in-session continuity; GenZAgents handles cross-provider + audit.
For enterprise use
For organisations: ChatGPT Memory at scale is a lock-in trap. The receipts approach is the cross-provider hedge. Most enterprise customers using ChatGPT for Work pair it with GenZAgents to maintain portability + audit.
Will OpenAI ship a portable Memory feature?
Unlikely. Same logic as Anthropic's memory: cross-provider portability hurts lock-in. They'll improve the in-ChatGPT memory, not the cross-provider story. The cross-provider layer has to live elsewhere; that's GenZAgents.