Google is introducing new memory import features for Gemini that can bring an understanding of users' key preferences, relationships and personal context directly into the platform. The features are rolling out to free and paid consumer Gemini accounts, adding to what is becoming a broader industry pattern of reducing friction when switching between AI assistants.
The tools work through two complementary approaches. Users can head to their settings and select the import option, where they'll get a suggested prompt to copy and paste into their current AI app, then copy that app's response and paste it back into Gemini. The system supports up to five .zip files per day, with a size limit of 5 gigabytes each for those wanting to import full chat archives.
Google is also renaming its "past chats" feature to "memory," a change rolling out in the app over the next few weeks. This terminology aligns with how competitors frame personal context, and Gemini's Personal Intelligence feature can now include relevant insights from Gmail, Photos and Search history in its responses when users grant access.
Following Anthropic's lead
Google's move follows Anthropic's launch of a memory import tool for Claude earlier in March, available to free users. Anthropic's tool allows users to import conversations and memories from other AI providers like ChatGPT using a prompt that Anthropic prepared for other chatbots. The basic mechanics are similar: a standardised prompt generates a summary from the competing platform, which users then paste into their destination app.
What's notable is how quickly these features have emerged. Google has been testing a similar import feature for Gemini since February 2026, but its approach imports full chat histories rather than distilled memories, and imported data is saved to Gemini Activity. Meanwhile, OpenAI has not announced an equivalent import feature, and as the market leader with 300 million weekly active users, lowering switching costs does not serve the same competitive interest.
Addressing the switching cost problem
The reason these features matter centres on what economists call switching costs. Most users have spent months or even years training a specific chatbot to understand their writing style, professional background and personal preferences, and leaving that built-up context behind feels like losing a digital memory. When that knowledge is locked inside ChatGPT or Claude, users feel reluctant to experiment elsewhere, even if they suspect a competitor might better suit their needs.
The new features are not available to users with business, enterprise or under-18 accounts, narrowing the rollout to general consumer users for now. Google aims to create a more cohesive assistant that combines imported insights with data from Google services like Gmail, Photos and Search, provided users have granted necessary permissions.
The broader pattern is clear: as the AI assistant market becomes more competitive and fragmented, platforms are discovering that user data portability is becoming a feature unto itself. Rather than trying to lock users in through information they cannot easily move, providers are building import tools as a way to attract switchers and lower the perceived risk of experimentation. Whether this trend continues, and whether users ultimately benefit from greater interoperability, will shape how the AI market develops over the coming year.