I don’t know if it’s possible to offer ideas for improvement here, but judging by the existence of the “Feature Request” tag, it’s possible
I am proposing numerous improvements to the web version of Google Gemini.
Integrating AI agent tools into the web version of Google Gemini
This is an opportunity to enhance programming with Google Gemini. I propose creating an AI sandbox where the AI can create and edit files, then send them directly to the user via chat or Google Workspace. It would be something like a /home/gemini folder where the AI has terminal access and can execute commands such as sudo apt update or npx vite.
Integrating skills into Gemini settings
In the Gemini “Personal Intelligence” settings, it would be possible to implement a feature allowing users to upload their own skills and invoke them using @skill-name, similar to how it works in Google Antigravity, but using a slash instead.
Increase the number of connected applications
In addition to Google Workspace and GitHub/Verify/Contacts apps, include applications such as Stripe/PayPal for transaction management, Shopify for store management, and Zillow/Redfin for real estate searches.
Projects system
You can add a Projects tab to the left-hand menu, alongside Notebooks and chats. A project functions as a group of chats with its own independent system prompt and context; when creating one, you specify the project name, description, and instructions for the AI. By combining this with skills, you can transform Gemini into a business assistant, where one skill acts as a Product Manager, another as a Programmer, and a third as a Design Docs Writer.
Tools for music creation
When creating music using Gemini/Google Flow, it is possible to implement editing tools—such as remixing, modifying specific sections, or removing vocals (to create an instrumental version of a song).