Google’s Gemini adds memory-import to ease switching between AI assistants

What’s new

Google (谷歌) has rolled out a memory-import feature for its AI assistant Gemini, a move aimed squarely at the pain point of platform switching: users can now migrate conversation histories and personal preferences so their new assistant “remembers” them from day one. It has been reported that Gemini supports two migration methods. One is a prompt-copy workflow where users copy a specified prompt into their old assistant, let that assistant summarise what it knows about the user, and then paste the summary back into Gemini. The other is a bulk import: users reportedly can export conversations from other providers, package them as a ZIP and upload—Gemini accepting files up to 5GB—to transfer long-term chat history wholesale.

Why it matters

This is a straightforward usability win. Changing AI assistants has typically forced users to re-teach preferences, work habits and unfinished projects. With memory import, Gemini can pick up a writing project, sustain a communication style, or retain calendar and preference settings without the friction. But there’s a trade-off. Google says imported conversations and subsequent interactions will be saved in activity records and used to optimise services and train models; users can reportedly manage or delete those records. Translation: easier continuity, but more data flowing into the company’s training pipelines.

Context and implications

For Western and global readers less familiar with China’s tech landscape: Google’s consumer services are largely inaccessible inside mainland China, where local players such as Baidu (百度), Alibaba (阿里巴巴) and Tencent (腾讯) dominate AI assistant development. Still, features that improve portability and lower switching costs matter globally—especially as governments and regulators scrutinise how AI firms collect, store and reuse personal data. Will regulators demand stricter controls on cross-platform memory transfers? And how will domestic Chinese platforms respond with their own migration tools, given different privacy regimes and data localisation rules? Those are the questions this update raises in a rapidly competitive, geopolitically charged AI market.