Ragie Connect: Build RAG Apps Over Users' Data
Reduce dev time from months → mins.
A ChatGPT user can add documents from Google Drive or OneDrive to do RAG over them:
This is a pretty handy feature.
But building the infra to support this can be a nightmare if you, as an application developer, want to integrate similar capabilities in your application for your users.
Today, let’s discuss Ragie Connect, an extremely powerful solution to build RAG apps on users’ data.
We’ll understand why building this end-to-end solution is challenging and how Ragie Connect solves this in minutes, along with a demo and implementation.
The challenge
Here’s what it will take to build the above infra in-house:
Set up OAuth flows for all integrations (Slack, Notion, Google Drive, OneDrive, Confluent, Jira, Salesforce, etc.)
Build UIs for user folder selection or other details based on the integration.
Maintain sync infrastructure to keep the data updated.
Handle indexing and partitioning between users efficiently.
So far, we have only been able to gather the data from your user.
We still need to build the RAG pipeline—chunking, indexing, retrieval, and generation.
As you may expect, this can be prohibitively expensive and time-consuming to build from scratch.
Ragie Connect solves this in minutes by providing the entire infra to handle authentication, authorization, sync process, and efficiently indexing users’ data:
This way, any software developer can build RAG-enabled AI apps in minutes without cultivating extensive AI expertise.
Ragie takes care of everything.
Let’s look at a demo of Ragie Connect.
Demo
Here, we have a locally running chatbot that was augmented with RAG capabilities in just 20 minutes (code is linked towards the end). It uses Ragie Connect to allow users to connect and chat with their knowledge base:
Once the user logs in, it shows an elegant user interface to chat with.
As a user, I connect a folder from my Google Drive, as shown below:
Done!
Finally, as a user, I can chat with this data:
Similarly, I can connect to any relevant source, while Ragie Connect will handle all backend integrations.
Implementation
As a developer, you only need to add a few things to your existing apps.
Define a Ragie client and add a redirect URL when users click “Add data.”
From here, Ragie takes over. It fetches the data and manages the end-to-end integration.
Once the user has connected the data, Ragie redirects the user back to the provided redirectUri.
Done!
That was simple, wasn’t it?
Conclusion
In the current state of AI, the effort that goes into building real-world RAG apps is vastly underestimated.
This is because developing RAG apps is an AI job AND a software development job.
In that respect, Ragie Connect converts a resource-intensive process into a seamless, developer-friendly experience.
Instead of spending weeks (or months) building the infrastructure to integrate user data, sync it, and make it usable for RAG, application developers can now achieve this in minutes, as we saw above.
Here are some examples:
Imagine a platform designed to help lawyers draft legal documents by analyzing case histories. With Ragie Connect:
Lawyers can seamlessly sync their Google Drive folders containing past case files.
The platform will automatically index these documents.
The platform developers can save months of work by leveraging Ragie Connect’s infra.
Imagine a platform for marketers to generate campaign strategies based on past performance. With Ragie Connect:
Marketers can sync analytics data.
The tool uses RAG to analyze trends, audience behavior, and past campaign outcomes, providing actionable insights in real-time.
Developers leverage Ragie Connect to support diverse integrations without worrying about maintenance or scalability.
And many many more use cases!
Find the code for the above demo here: Basechat GitHub.
Thanks to Ragie for showing us their powerful RAG infra capabilities and partnering with us on today's newsletter.
Thanks for reading!