Google is adding Nano Banana-powered image generation to Gemini Personal Intelligence.
How the personalised image tool works
Google says Gemini’s Personal Intelligence will now be able to generate images that draw on a user’s own account context. The system uses the Nano Banana image model to turn short prompts into pictures that reflect information already linked to a person’s Google account, such as items in Gmail and labels in Google Photos. You don't have to list your interests or group labels in a prompt — Gemini will use your account context to reflect them in images.
Try typing "Design my dream home" — Gemini will shape the result using the profile data tied to your account. Google Photos labels are part of the connection. So when a user asks for an image of “my family and me doing our favourite activity,” Gemini can rely on labeled groups like “Family” to produce people and settings that match the user’s context.
Gemini's chat interface now taps the Nano Banana model so images can reflect what Google already knows about you, not just generic stock-like pictures. Google says the Personal Intelligence feature will display a new “sources” button that shows where Gemini pulled contextual cues from when creating an image.
Controls, feedback and reference images
Google warns the context link isn't perfect. The company notes users will be able to provide feedback when Gemini applies context incorrectly. That feedback is intended to help refine future outputs and reduce errors in how the model reads account data.
Users can also add reference photos to guide image generation. A “+” icon in the image interface lets a user attach pictures that the model can use as visual anchors. That gives people more control over the look and composition of generated images while still allowing Gemini to factor in known preferences and labels.
Google says you'll manage these settings the same way you do other account connections — through your Google account's consent and permissions screens. The company says the feature will behave like other account connections: users can expect the usual settings to view and manage what Gemini can access.
Rolling out to subscribers first
The new image-generation feature will be available to Plus, Pro and Ultra subscribers in the United States within the coming days, Google says. The company plans to add the capability to Gemini in Chrome on desktop and to expand access to other users over time.
Personal Intelligence itself launched earlier in the year and reached all U.S. Users in March. Google has been widening the feature’s footprint internationally; this week the company extended Personal Intelligence access to more users in markets such as India and Japan, and the image-generation integration is the next step in the product’s push toward broader availability.
Why Google is pairing Nano Banana with Gemini
Nano Banana is an image model Google uses to generate both photorealistic and stylized visuals from text prompts. Google’s move ties a model that’s optimised for image synthesis to a conversational assistant that already holds cross-product context about an individual user. That pairing aims to cut the friction of prompt engineering — fewer details required from the user, faster generation of images that feel personalised.
For Google, this gives Gemini a point of difference: it can generate personalised images from shorter prompts. Many competing assistants and image tools require users to include explicit descriptors in prompts if they want results that reflect particular people, places or interests. By leveraging a user’s account metadata, Gemini can make shorter prompts work harder.
But tying personal account data to image generation raises clear safety and design challenges. Models that draw on personal data need guardrails so outputs don’t reveal or mishandle sensitive information. Google’s mention of a sources button and feedback mechanism suggests the company is building transparency and correction paths into the flow, but how those features operate in practice will matter for adoption.
Implications for creators and everyday users
For creators, the ability to combine reference photos with context-aware synthesis could speed up workflows. Illustrators, designers and social creators may use Gemini to iterate visual concepts that reflect a client’s style or a user’s documented preferences. The feature could reduce back-and-forth on briefs and let teams generate rough layouts or mood boards quickly.
Everyday users may find the tool helpful for personal projects — event invites, family mock-ups, or visualising plans like a home redesign. Gemini’s use of Google Photos labels could make it easier to summon images that include specific people or groups without having to type names, as long as those people are organised into labelled collections in a user’s library.
But there are limits. Google warns Gemini may get context wrong. The company expects users to review sources and, if needed, correct the assistant. That requirement suggests generated images shouldn't be treated as definitive representations of people or events without user verification.
How this fits into Google’s broader AI strategy
Adding personalised image generation to Gemini follows a pattern: Google is building experiences that blend generative models with the company’s existing data and services. By tying models to account context, Google hopes to deliver more useful, less generic outputs across writing, summarisation and now image creation.
Earlier expansions of Personal Intelligence gave Gemini the ability to summarise emails, highlight calendar events and remember user preferences. The new image capability is a natural extension of that memory-like feature set — one that moves from text to visuals while using the same account-linked signals.
That strategy is likely to shape how competitors approach assistant-driven creativity. Companies that lack the same depth of account-linked signals will need other ways to let users supply personal context, such as richer on-device memories or simpler reference workflows.
What to watch next
Key questions include how quickly Google rolls the feature beyond U.S. Subscribers, how transparent the sources button proves to be, and how reliably Gemini pairs Nano Banana outputs with correct account context. Those factors will determine whether the feature becomes widely useful or remains a niche convenience for subscribers who already trust Google with large amounts of personal data.
Google’s rollout schedule suggests the company will test the feature with paying users before pushing it broadly. Early feedback from those subscribers is likely to shape tweaks to the interface, the feedback tools and the guardrails around sensitive content.
For now, the addition shows how large AI systems are moving from generic generation toward outputs that reflect the person using them.
Related Articles
The image-generation feature will be available to Plus, Pro and Ultra subscribers in the U.S. Within the coming days.
This article was created with AI assistance.
