Case Studies
Connect external data to LLMs, no matter the source.
"For months we were building our own web scraping solutions, which to be honest, was a major distraction, plus, the thought of having to build integrations with things like Google Drive, Notion, Zendesk etc brought us out in hives. Then we found Carbon and life has been so much easier since! We’ve been able to focus on what we do best and leave the non-core bits of our value proposition to them. Add to that the fact that they are insanely responsive and reactive to bugs and issues and we couldn’t ask for a better partner to work with to build My AskAI. We’re looking forward to adding more of their functionality into our products in the future (if we can keep up with their development speed!)."
Mike Heap
Co-Founder, My AskAI
Challenge
My AskAI has more than 45,000 users who utilize their product to create AI Assistants. These AI Assistants are primarily used as user-facing customer support assistants on company websites. In addition, they also have products that enable businesses to create AI internal knowledge assistants to retrieve information and provide answers directly back to users in messaging platforms such as Slack, Teams, and WhatsApp. My AskAI needed to integrate various data sources, such as Google Drive, Notion, and website content, into their retrieval pipeline to provide comprehensive and accurate answers to user queries.
Solution
When we first started working with Mike and Alex, we helped them easily integrate Google Drive and Notion into their retrieval pipeline. We iterated fast and incorporated a lot of their feedback to improve the end-user experience for all customers. This led to the launch of their conversational AI assistant for Slack, which could answer internal questions based on the company's docs and wikis in Google Drive and Notion.
The launch was a success, and, a month later, the My AskAI team came back and wanted to pilot our in-house web scraper as well. They wanted the ability to quickly scrape up to 2,500 pages recursively from either URLs or sitemaps, manage scheduled re-syncs, and handle Javascript-heavy websites. Our web scraper ended up being the perfect solution for their need, and the fact it was cost-efficient made it a no-brainer to switch over. In December My AskAI launched their 2.0 product that allowed customers to create AI Customer Service assistants, trained on your company's website and help docs. In addition to the web scraper, My AskAI leverages Carbon's infrastructure to vectorize the scraped web content and retrieve relevant data using our blazing-fast semantic search (average <0.2 latency).
Conclusion
We're excited to continue growing our partnership with My AskAI. Alex and Mike have been a pleasure to work with and their feedback has been incredibly helpful in improving the Carbon developer experience for all. We are actively improving our hybrid search capabilities to enable My AskAI customers to search through all web pages and find relevant content.
