Case Studies
Connect external data to LLMs, no matter the source.
"Having to build and maintain a data extraction, embedding generation and retrieval system on our own was a major resource consumer and a big distraction until we discovered Carbon. Once we were fully integrated with them we simply tripled our speed and now we can focus on the thing that matters, the app itself. Secondly, Carbon is GDPR compliant and super secure, taking another major stress off of our shoulders. They truly have been a partner and responded very quickly to any issue we had, including pushing features when we needed them. In an AI era I believe that Carbon should be a must in anyone’s tech stack."
Vlad Racore
Co-Founder, Research Studio
Challenge
UX Researchers and Product Designers utilize Research Studio to convert both qualitative and quantitative data collected from users during research into actionable insights using Generative AI.
Before using Carbon, researchers and designers stored research data in various file formats across multiple data sources. They had to download all these files locally and individually upload them to Research Studio. After receiving feedback from users that their previous flow introduced unnecessary friction, Research Studio evaluated options to streamline file ingestion for a better customer experience. After evaluating several solutions, Research Studio decided to work with Carbon because of our ease of implementation and support for connectors popular among researchers and designers.
Solution
We worked with Research Studio to integrate our Carbon Connect component, which provides a pre-built user interface for users to authenticate and upload content from cloud storage services like Google Drive, OneDrive, Dropbox, and more. It also has screens for uploading local files and submitting URLs for web scraping. The component is entirely customizable, allowing developers to enable multiple data sources, file formats, processing settings, and branding options in minutes.
Once Carbon processes the user files, Research Studio uses our built-in high-speed hybrid search to extract relevant content from the uploaded documents. Hybrid search allows Research Studio to effectively search through files containing tabular data such as CSV, XLSX, and PDFs. The search results serve as training data for the Language Learning Models (LLMs) available in Research Studio. Furthermore, Carbon's automatic refresh feature ensures that the search is updated whenever users add new files or update existing ones with new content, providing an added benefit.
Conclusion
We have established a great working relationship with the team at Research Studio and are continually finding new opportunities to expand our collaboration. Our current focus is on having Carbon provide more web scraping infrastructure for SiteGPT. Additionally, we intend to integrate our hybrid search with SiteGPT's AI chatbots, allowing them to more accurately search through all files and data sources to find relevant content.
