3DN Respond PDF Knowledge Base Integration Tutorial
Walks through uploading PDFs to the knowledge base, including chunking, embedding, and semantic search capabilities.
Transcript:
We built 3D and respond to transform and customer into exceptional service experiences, and we do this by streamlining, prioritizing and resolving requests across all communication channels. And this empowers your team to deliver faster, more personalized support that builds lasting relationships. Today I wanted to show you how you can add to the knowledge base in 3D and respond. Previously we looked at using the web crawler. Today let me walk you through how 3D and respond processes PDF documents and uses them to power intelligent customer responses. It’s important to note that you’ve got a lot of information in your business, some of which is available on the website. But there’s a lot of information that’s not on your website too. And so this gives you an opportunity to add that knowledge directly into 3D and respond. When you upload a PDF into 3D and respond we do much more than just store the file. We actually read the document using specialized processing technology and this extracts both the text content and metadata like author creation date, page count, that sort of thing. However a full PDF document is way too much information for the AI to process all at once. So we break each document down into smaller digestible chunks and we add some overlap between these chunks to maintain context and continuity. And here’s where it gets really interesting. Each of these chunks is transformed into a mathematical representation. This is essentially a set of numbers that captures the meaning of the text in a way that computers can understand. It’s like translating the text into the AI’s native language. This information is then stored in a special database that can perform semantics searches, finding content based on meaning, not just keywords. And this is far more powerful than traditional search methods. When a customer inquiry comes in, 3D and respond transforms the question into an optimized search query. It then searches across all of your knowledge based documents, finding the most relevant information based on its actual meaning. The search result become part of the context that helps the AI understand how to respond. They’re included when the system classifies the customers intent. And when it generates a response, this ensures that responses are not just conversationaly fluent but also factually accurate and based on your official documentation. For example, the customer asks about your refund policy for an event. The system can pull the exact details from your PDF documentation, ensuring the response matches your official policy perfectly. This knowledge-powered approach means that your customer service team doesn’t need to memorize every detail. The system automatically accesses the right information at the right time, making responses both accurate and consistent.