Earlier this month Marissa Mayer, the former Yahoo CEO, said she was ready to introduce Dazzle, a personal AI assistant that raised an $8 million seed round last December. Unlike many recent assistants that rely heavily on text-based apps such as email, calendars or shopping history, Dazzle derives its user context primarily from the photos stored on a person’s phone.
Photos as a source of personal context
"I think that photos are an underappreciated source of information," Mayer said. Dazzle claims that by analyzing a user’s camera roll it can infer hobbies, interests, food and style preferences, how someone likes to spend time, and who they spend it with. Mayer likened the value of an entire photo library to millions of words of context.
Two main experiences
In Mayer’s demo, Dazzle’s functionality is split into two core experiences:
- Immediate tasks: scanning recent photos to extract actionable details — for example, creating a calendar entry from an event flyer or finding a repair person after spotting a broken garage door in a photo.
- Long-term personalization: mining the entire photo library to generate tailored suggestions for vacations, gifts and activities.
Examples from the demo and early use
Mayer says she uses Dazzle to brainstorm family outings. By scanning her photos, the assistant inferred the family’s interest in escape rooms and suggested several San Francisco Bay Area locations she hadn’t previously known about. When tested for vacation recommendations, Dazzle suggested several Mediterranean destinations, apparently picking up on past visits to Spain and Greece, and surprisingly included Sicily, which the writer had visited four years earlier.
The system is not without blind spots: it failed to remember that the writer’s daughter already knows how to roller skate when asked whether to buy her a pair for an upcoming birthday. Nevertheless, the assistant offered compelling, personal suggestions such as a local pottery studio or a bioluminescent kayaking tour in Tomales Bay.
Background and prior projects
Mayer’s earlier startup Sunshine launched an AI-powered photo-sharing product called Shine in 2024. Shine was criticized for an outdated design, did not achieve broad adoption and was eventually shut down, but Mayer says the project produced “interesting IP” that informed Dazzle’s development.
Privacy stance and the competitive landscape
Several new AI assistants have appeared recently — including Meta’s Muse and Instinct — some of which have sparked security concerns. Mayer argues that users might be more comfortable granting an AI access to their photo libraries than to highly sensitive data like emails and messages. She emphasized that Dazzle prioritizes privacy and discards any personal information the AI flags as sensitive.
What this means for users
This wave of personal AI assistants presents new options for consumers. While Dazzle may not yet be as broadly useful as some competitors, it demonstrates a path where AI does more than execute commands: it can build a personal model of who you are from images. Mayer and her team expect the product to improve over time as they refine capabilities and address gaps in memory and understanding.
