Behind the Scenes: Decoupling Integrations at everfind.ai
02. February 2025Welcome to an early peek into the work-in-progress behind everfind.ai (formerly refind.ai). Today, I want to share some insights into our integration architecture—a critical part of making our platform both flexible and secure.
The Integration Translation API
At everfind.ai, we’re pioneering an architecture that decouples our core backend from external integrations. The key to this approach is our integration translation API, which serves as a bridge between our backend and a specialized translation layer.
This translation layer is designed to handle all the nitty-gritty details of interfacing with various integrations. By abstracting these responsibilities away from our backend, we ensure that the system remains modular, maintainable, and scalable.
Establishing Trust with .well-known Endpoints and JWKS
To facilitate seamless communication between our backend and the translation layer, we’re leveraging a couple of industry-standard approaches:
- .well-known Endpoints: These endpoints simplify service discovery and ensure that both layers can locate each other reliably.
- JWKS Request Signing: By signing requests with JSON Web Key Sets (JWKS), both our backend and translation layer can authenticate and trust the requests they exchange.
This combination of discovery and security protocols forms the backbone of our trusted communication channel.
The " Adapter Way" for Custom Integrations
One of the most exciting aspects of this design is the freedom it offers developers. With our decoupled architecture, users can now write custom integrations that connect seamlessly with everfind.ai. Whether you’re integrating a new tool or connecting a 3LE OAuth2 app, you can do it the “adapter way.”
In practice, this means that you can develop an adapter—an intermediary component—that translates third-party OAuth2 protocols into a format that everfind.ai can understand. This not only enhances flexibility but also empowers our community to extend our platform in ways that best meet their unique needs.
What's Next?
It’s important to note that everfind.ai is still in development. The work on these integrations is ongoing, and there’s plenty more to come. Our goal is to create a robust and extensible system that allows for rapid iteration and growth, all while maintaining a high level of security and reliability.
Stay tuned for future updates where I’ll dive deeper into the technical details and share more about our journey building everfind.ai.