[DigitalToday reporter Yoonseo Lee (이윤서)] Google Gemini's custom feature 'Gems' is drawing attention as a personalization tool that reduces repetitive prompts.

Tech publication TechRadar reported on July 7 (local time) that Gems, similar to ChatGPT’s custom GPTs, remembers specific roles, preferences and conversation history. It helps users avoid having to explain the same context each time they start a new chat. Users can set up dedicated assistants by purpose and call them up as needed.

A 'family trip planner' made for family outings remembers conditions such as a child's age, travel distance, stroller use and preferred places, and reflects them in weekend itinerary suggestions. With the Deep Research feature added, it checked whether venues were open that weekend, seasonal events, hours and admission fees before planning. If rain was forecast, it switched the plan to focus on indoor children's museums and indoor play spaces. After a condition was added that long restaurant wait times make it hard to move with young children, it later prioritized simple cafes or picnic spots.

A hobby 'hobby coach' was designed for activities that are difficult to restart, such as guitar or stargazing. It remembers the user's current skill level, equipment, goals, schedule and learning style, then continues from previous conversations. For guitar practice, it suggested progressively more difficult chord exercises and song recommendations and set goals in multi-week increments. For stargazing, it suggested different observation targets depending on the season and the moon phase.

A content recommendation 'movie and TV curator' remembers streaming subscription status, preferred genres, what the user has already watched and the family members watching together. Reflecting the condition of having young children at home, it narrowed recommendations to low-pressure family movies rather than focusing on popular titles. Adding a prompt such as, "Do not repeat recently recommended titles and explain why they fit the user's preferences," can further improve recommendation quality.

Image-based uses were also presented. An 'outfit planning chart' learns a user's usual style, color preferences and frequently worn outfits, then creates styling drafts based on photos. It was also used to check virtual outfits before buying for theme parties, Halloween and travel outfits, as well as everyday wear. This Gem can provide 'realistic style visualization' based on uploaded photos of clothing, accessories or the user.

The article also included a case combined with a music generation tool. A 'personal theme song' uses a music tool recently added to Gemini to create an original song from a simple theme. It first learns the user's preferred music style and mood, then makes a song through questions about a person, pet, place or memories.

Gems focuses on maintaining continuity of a role, unlike saved prompts. It remembers a user's work style and preferences, and can be expanded to tasks where repeated context matters, such as meal planning or packing for travel. The structure, which carries over roles and preferences without repeating the same instructions, is drawing attention as an attempt to turn AI from a one-off tool into a task-specific interface.

If you’re not already using Gems to optimize your workflow in Gemini, it’s time to start. Gems allow you to quickly reuse a prompt and add reference files. Open the side panel, create a gem, and turn your repetitive tasks into a single click. pic.twitter.com/m3zya8LUM1