Google is making Gemini a little more “you.” The tech giant has started rolling out a Personal Context feature for Gemini 2.5 Pro, letting the chatbot remember past conversations and tailor responses—something OpenAI and Anthropic have offered for months. But there’s a catch: Google’s approach is cautious, with limited controls, no editing of preferences, and slower feature expansion.
Key Takeaways
- Personal Context lets Gemini remember past chats and tailor answers.
- Rolling out first to Gemini 2.5 Pro in select countries, coming soon to 2.5 Flash.
- Unlike rivals, Google still doesn’t let users edit or delete preferences.
- Temporary Chat and new data controls aim to improve privacy.
- Google trails OpenAI and Anthropic in full memory features.
Google’s Slow Move Into Personalization
For years, chatbots have struggled with one thing: remembering who you are and what you’ve already told them. OpenAI’s ChatGPT and Anthropic’s Claude have been making strides here, letting their models reference past chats automatically. Now, Google is finally making its move.
The company announced a gradual rollout of Personal Context—a feature that will “learn from your past conversations and provide relevant and tailored responses” without you having to constantly remind it of your preferences. This is part of Google’s vision to make Gemini “an AI assistant that truly understands you,” said Michael Siliski, Senior Director of Product Management.
How It Works—and What’s Missing
Until now, Gemini users had to manually refer to old conversations if they wanted the model to use prior details. With Personal Context enabled, Gemini will do this automatically for things like your project details, preferred writing style, or company branding.
But there’s a limitation: unlike ChatGPT and Claude, Gemini doesn’t let you edit or delete stored preferences yet. You can turn Personal Context off entirely, but you can’t fine-tune what it remembers.
The feature is available now for Gemini 2.5 Pro in select countries and will expand to Gemini 2.5 Flash “in the next few weeks,” Google said.
Privacy and Temporary Chat
To address privacy concerns, Google is adding Temporary Chat, similar to a feature OpenAI rolled out in April 2024. These one-off chats won’t influence future responses and won’t be used to train Google’s AI models.
Additionally, Google is introducing more granular data controls. A new toggle lets users prevent their conversations from being used in future model training. If enabled, only a small sample of future uploads might be used to improve services—otherwise, users can keep their data completely private.
For those who prefer minimal data storage, Gemini already allows chats to auto-delete after 3, 18, or 36 months, and can store chats for only 72 hours if activity saving is disabled.
Why Google Is Playing Catch-Up
This rollout comes a full year after OpenAI and Anthropic launched similar personalization features.
- OpenAI’s ChatGPT has had memory since 2024, recently updated to reference all past conversations automatically.
- Anthropic’s Claude introduced “Styles” last year, letting users customize tone and interaction style, and this week enabled full conversation referencing without user prompts.
By comparison, Google’s earlier Gemini versions could only recall past information if directly prompted—and its new update still lacks the fine control power users have with rivals.
Why This Matters for Users
For individual users, personalization means less repetition and more relevant answers. For enterprise teams, it’s even more important—AI assistants need to remember brand guidelines, project history, and voice without re-explaining every time.
The lack of editing options in Google’s rollout might frustrate power users, but the introduction of Personal Context and Temporary Chat is a step toward a more helpful and privacy-conscious assistant.
Conclusion
Google is late to the personalization party, but it’s finally joining. The new Personal Context feature in Gemini 2.5 Pro is a promising start, especially paired with Temporary Chat and better data controls. Still, without full memory editing and deletion, it’s clear Google is prioritizing caution over speed—a choice that could cost it ground against Anthropic and OpenAI in the AI assistant race.