Google Launches Personalized Gemini Service in Taiwan With Gmail and Photos Integration

Google introduced Gemini Spark at its I/O developer conference in May 2026 as a 24/7 personal AI agent powered by Gemini 3.5 and built on its Antigravity architecture. Unlike a conventional chatbot, Spark runs in the cloud and can continue scheduled tasks after a laptop is closed or a phone is locked. Its links to Personal Intelligence, Gmail and Google Workspace let it organize inboxes, monitor news and compile research, underscoring Google’s push from AI answers toward software that can act for users.

Google said on July 14 that Spark was rolling out in all Gemini Apps languages to Google AI Ultra subscribers wherever the service is supported, including Taiwan, but excluding the European Economic Area, Britain, Switzerland and Nigeria. On July 16, the company began extending the English-language service to Google AI Pro subscribers in the United States. The update also added Workspace actions, including editing private spreadsheets and presentations, reading comments and revising shared documents, while keeping sensitive actions such as purchases or sending emails subject to user approval.

Google introduced Gemini’s Personal Intelligence in the U.S. on Jan. 14, 2026, allowing users to opt in and connect Gmail and other Google services. The system can retrieve details across emails, including merchants, transaction dates and receipt amounts. That capability turns scattered renewal notices into a personal spending record, making it easier to identify recurring charges that may otherwise remain unnoticed. Google says Gmail inboxes are not used directly to train its models.

Google began rolling out Personal Intelligence in Taiwan on April 14, initially for Google AI Plus, Pro and Ultra subscribers, with free users due to follow. After connecting Gmail, users can ask Gemini to find subscription receipts and, in about one minute, produce a list of services, individual charges, renewal dates and total spending. They can also ask it to flag subscriptions for review before the next debit. No standard amount applies because the calculation is drawn from each user’s own emails.

Since consolidating Bard and Duet AI under the Gemini brand in 2024, Google has continued embedding generative AI into Android rather than keeping it as a standalone chatbot. Gemini Intelligence can understand on-screen content, voice commands and context across apps, carrying out multistep tasks such as booking restaurants, hailing rides and completing forms. The move positions AI as a core operating-system capability.

Google unveiled Gemini Intelligence on May 12, 2026, with features including cross-app automation, webpage summaries in Chrome, writing refinement in Gboard and natural-language widget creation. The service will first arrive on the latest Samsung Galaxy and Google Pixel phones in summer 2026, before expanding to watches, cars, glasses and laptops by year-end.

Generative AI assistants accumulate users’ preferences, relationships and work context over time, but those personalized settings and conversations are often difficult to preserve when switching platforms. Google has added data-transfer tools to Gemini to lower the cost of moving from services such as ChatGPT, extending competition among AI assistants beyond model capabilities to data portability.

Google announced on March 26, 2026, that it had begun rolling out memory and conversation import features on Gemini’s web version to consumer accounts on both free and paid plans. Users can paste in preference summaries generated by other AI services or upload .zip archives of conversations up to 5GB, then search and continue past chats. The features are not yet available for enterprise, business or under-18 accounts.

Google began testing Personal Intelligence with Google AI Pro and Ultra subscribers in the United States in January 2026. With users' consent, the service connects to Gmail, Google Photos, YouTube and Search to provide context-aware help with shopping, troubleshooting and travel planning. The move signals a shift in AI competition from general-purpose answers toward assistants that work across services, bringing data permissions and privacy controls into sharper focus.

Google announced on March 17, 2026, that Personal Intelligence was expanding to free-tier accounts in the United States. It became available in Search AI Mode that day, with phased rollouts planned for the Gemini app and Chrome. Users can enable or disable connections to Gmail and Google Photos. Google said it does not directly train its models on inbox or photo-library content, and Workspace accounts are not eligible.