Announcing the Crescendo knowledge base feature

The best way to ensure accurate answers from a generative AI application is to utilize a high quality base of Knowledge. The Crescendo Knowledge Base consists of files containing all the information our CX Assistant needs to converse with customers – everything from a company’s blog posts to its product FAQs. It’s a repository of all the content a customer could find useful on a company’s website, and it’s the material we use to train our bots.

Each company we partner with has its own, unique knowledge base that we create to provide content for our Crescendo CX Assistant. For each company, we “scrape” their websites to collect the knowledge content for their specialized knowledge base. When a customer asks a specific question about a company’s product offerings, we use a process called Retrieval Augmented Generation (RAG) that allows our CX assistant to search a company’s knowledge base and deliver the relevant information to the customer.

Here is a representation of the Crescendo Knowledge Base that’s used by our CX Assistant to answer questions about Crescendo – it contains several documents that delineate answers to common questions and descriptions of our company and services:

An important benefit of the Knowledge Base is preventing “hallucinations”, or mistakes made by LLMs when generating responses. Because general purpose LLMs don’t always have the right context for answering a user’s questions, they can create false or misleading responses. Incorporating the Knowledge Base allows us to avoid this issue by constraining our CX Assistant to only reference information in the knowledge content. That way, our bot delivers fact-based answers with proper context rather than coming up with its own conclusions.

Sometimes, we receive the wrong answer from our CX Assistant. In these cases, we know that there is an underlying issue with the knowledge content. Fortunately, it is quite easy for us to identify where the wrong information came from – the transcripts on our c7o.ai platform contain the sources from the knowledge base our bot used for each response. We can then edit the knowledge content to reflect accurate information and inform our clients of any errors in their published content.

Most importantly, our Knowledge Base can be easily updated. As companies rapidly iterate with changes in service pricing and new product releases, we reflect those changes by editing existing files in the company’s knowledge base or uploading new knowledge content. Our CX Assistant can immediately start referencing new information once the knowledge base is updated. This ensures that customers receive accurate and up to date information about any changes in service or product offerings.

In the following video, we demonstrate the process for updating our knowledge content. When we notice that our bot fails to deliver a satisfactory answer to a customer, we add the right information to our knowledge base. Notice how simple it is to modify our knowledge content:

As we introduce the Crescendo Knowledge Base feature, it is clear that the real power behind our bot comes from strong knowledge content. It’s what allows us to keep customers happy when they receive quick and accurate responses. Through our knowledge base, we can easily automate and deliver high-quality responses to users. This commitment to efficient solutions is what makes Crescendo a leader in delivering cutting edge customer experience technology.

In conclusion, the Crescendo Knowledge Base is crucial for delivering consistent, high-quality responses via our Crescendo CX Assistant. By leveraging specialized knowledge we ensure our bot provides accurate and relevant information, reducing errors and enhancing customer satisfaction. At Crescendo, our commitment to excellence and innovation in customer experience technology is embodied in our robust knowledge base, setting the industry standard for service and support based on generative AI.