In AI applications, context often matters more than prompts.
I’m currently developing a universal and extensible platform fully integrated with AI.
The architecture is built around three layers:
• Application Layer: modular plug-in components used to build custom dashboards that can be stored either in a database or as JSON files.
• Database Layer: a universal access layer built on top of provider-specific implementations, featuring compiled SQL and specialized components.
• AI Layer: following the same approach as the database layer, with a universal abstraction layer and provider-specific implementations.
The context, memory, skills—or whatever terminology each vendor prefers—is provided by the components themselves. It remains fully under the control of both the application and the user, and can be stored in a database and reused across different tasks.
This first video provides a high-level overview of the architecture. Feedback and discussion are welcome.