TL;DR

Google is rolling out migration tools for its Gemini app that let users bring their conversation history and personal context from other AI assistants. The process uses a prompt-based transfer for memories and ZIP file uploads for full chat archives.

How the switching process works

The memory import follows a somewhat manual workflow. Users navigate to Gemini’s settings, select the import option, and receive a suggested prompt. They then paste that prompt into their current AI assistant, which generates a summary of stored preferences, relationships, and personal context. That summary is copied back into Gemini to seed its memory system.

For users wanting their full conversation archive, Gemini now accepts ZIP file uploads containing exported chat histories from other platforms. Google is also renaming “past chats” to “memory” across the app — a terminology shift that reflects how these systems increasingly function less like chatbots and more like personal knowledge stores.

The broader “Personal Intelligence” play

The switching tools are part of Google’s wider push into what it calls “Personal Intelligence.” This feature draws on a user’s Gmail inbox, Google Photos library, and Search history to build a more complete picture of who they are. The ambition is clear: Gemini should know you well enough to provide genuinely personalised responses, not generic ones.

This approach gives Google a structural advantage over competitors. No other AI assistant maker has access to the same depth of personal data across email, photos, and search behaviour — at least not without users actively feeding it in.

Looking forward

The move reads as both an offensive and defensive play. Google is making it easier to leave competitors while simultaneously building a moat around its own ecosystem. For UK users, the data implications are worth noting: funnelling Gmail, Photos, and Search data into an AI assistant raises questions about data protection that the ICO may eventually want to examine, particularly around purpose limitation under UK GDPR. The convenience is real, but so is the concentration of personal data in a single AI system.