Building OM AI: A Multi-Provider AI Workspace with Conversation Continuity

Gemini API
1

Hello Google AI Developer Community,

I am an independent student developer from India, and I have started building an early-stage Android project called OM AI.

OM AI is a concept for a unified AI workspace designed around one conversation system, user-controlled context and memory, and multiple officially supported AI providers.

The core idea is not to create a new foundation model.

Instead, OM AI would act as a layer between the user and officially supported AI services.

The vision is:

One Account
One Conversation History
One Memory System
Multiple AI Engines

For example, a user may interact with different officially supported AI providers through a unified workspace.

The important concept is conversation continuity.

If a user chooses to change an AI provider, the conversation itself should not need to start from zero. OM AI would manage the user’s conversation history and selected context, then provide the relevant context to the next officially connected provider through supported APIs or authorization methods.

The system would be designed to respect:

  • Official APIs and supported integrations
  • Provider policies and platform limitations
  • User privacy and user-controlled data
  • Authentication and authorization requirements
  • Provider-specific quotas and billing rules

I am not trying to bypass consumer subscriptions, quotas, or access restrictions.

The goal is interoperability through officially supported developer platforms.

I am particularly interested in understanding the best architecture for integrating Gemini into this type of multi-provider AI workspace.

Some questions I would appreciate guidance on are:

  1. What is the recommended architecture for securely integrating Gemini API into a multi-provider application?

  2. What are the recommended approaches for managing conversation context when a user switches between different AI providers?

  3. Are there recommended best practices for separating provider-specific conversation state from an application’s own conversation and memory system?

  4. What security and privacy considerations should an independent developer prioritize when handling user context and AI provider integrations?

In the future, I would also like to explore a real-time voice assistant component called OM AI Live, designed around officially supported real-time AI capabilities.

I have already started development of OM AI and am currently building the Android application and its core workspace architecture.

I would sincerely appreciate technical feedback, architectural suggestions, and guidance from the Google AI developer community.

Thank you for taking the time to read about my project.

— Independent Student Developer from India
Currently building OM AI

2

I want to talk with technical team or developer team of Google and share my idea I want to talk with technical team or developer team of Google, share and explain my idea