What exists.
What is still ahead.

Capture, revisit, and deliberate reuse.

• Consent-based capture of eligible, displayed user text on supported ChatGPT pages; capture can be paused.
• Project and conversation navigation, continuous reading, working-text editing and lexical search. Editing and retrieval UX are still being refined.
• Cross-conversation topic reading and optional AI organization grounded in source material. AI output needs review.
• Explicit material selection from inputs or topics, context preparation and preview, then copy or export. Changed material invalidates an old preview.
• Implemented paths for open export, segmented backup and staged restore. See verification limits below.

Engineering tests are not real-world proof.

Remaining external evidence includes live ChatGPT journeys, physical devices, signed distribution and same-identity updates, and private official history exports. Large synthetic datasets have been used in engineering tests; that is not proof that your real archive has been safely restored.

We do not currently promise one-click import of all history, universal source compatibility, unlimited capacity, a ten-year lossless-storage guarantee, or immediate public-store installation.

More ways to keep thinking.
No borrowing from the future.

A real external AI reader connector, stronger semantic retrieval, more sources of personal expression, quick writing and voice on mobile, and separately authorized cross-device capabilities are future directions. Website illustrations should not imply that these are already fully available.