CYDAS PEOPLE takes Momento Cache to prod for their ChatGPT AI-powered chatbot in just two hours

Industry: SaaS, Chatbot, AI

Use Case: Chatbot Cache

‍

Founded in 2020, CYDAS PEOPLE is one of the leading providers of HR tech services in Japan. We provide tools to help HR teams scale organizational management and employee services. Companies of all sizes—from hundreds of employees to tens of thousands—use our products to drive stronger employee engagement.

‍

Relieving overloaded HR teams

When employees contact HR with questions, it is imperative to answer quickly. The answers to these questions are frequently found within the multitude of FAQs and existing docs at the company, but they are often lost due to employees using different lexical terms in their searches. A query for "vacation days" will likely yield an exact match — but as soon as someone frames the query as "days off," the simple full-text query response may not put the vacation policy page at the top of the search results. Manual inquiries—no matter how easy they are to answer—quickly add up to create unsustainable workload for strapped HR teams.

This is why we set out to make an intuitive chatbot leveraging the power of large language models (LLMs): CYDAS PEOPLE Copilot Chat.

‍

Building an AI-powered chatbot

We were particularly excited about LangChain and the multitude of chatbots the developer communities have been building with it. There was just one problem: we wanted to temporarily store the chat history without setting up servers or paying for idle infra—and get high availability out of the box. That’s where Momento Cache came in—a completely serverless caching service. The experience was incredible: we were able to integrate with Momento to store chat histories in production in under two hours.

Shortly after we built our chatbot and started deploying it, we noticed that LangChain announced an integration with Momento. With that, we can now use Momento as a prompt cache in addition to a chat histories store. If a user asks a question that has already been answered, the chatbot will be able to provide the answer much faster without making any calls to Pinecone or OpenAI.

Ready for what’s next

We’re very happy with Momento. The service is easy to use, fast, and stable—and there are no maintenance windows. In the future, we plan to use Momento Cache for a much larger workflow: replacing the application cache for our main CYDAS PEOPLE products.

‍

Unlock cost savings and performance improvements for your chat applications with Momento Cache. Sign up and try it today.

‍