Can I Feed PDF Files To My Chatbot ?

The short answer is no, you won't be able to take a PDF and give it to a regular AI Engine chatbot to use. That being said, what you care about is the content of the PDF file, not the file itself.

There are multiple ways to make your chatbot aware of this content. If the data from the file is readable in one of your posts, then you can enable the "content awareness" feature and the AI Engine will automatically change the context of your chatbot by adding this text data so your chatbot will be able to use it.

If you have a vast collection of content that you want to efficiently index and search, utilizing embeddings with Pinecone can be advantageous. Embeddings can capture semantic similarities between documents, allowing you to perform similarity searches or retrieve relevant content based on user queries. In this case, your chatbot will be able to effectively navigate through your knowledge corpus when presented with questions related to these topics. You can test its effectiveness for your specific case directly in the AI Engine, as demonstrated here: AI Engine - Manual Embeddings.

If you want to establish embeddings, you can refer to this documentation for guidance. This will enable you to provide plain text data about your website and even generate this data automatically using an AI Engine that scrapes all your posts and pages. In your case, you might want to manually copy and paste the content of your PDF file inside the embedding content.

If you possess technical knowledge, you also have the option to host various PDFs on your server or any other location. You can leverage the AI Engine filters to query this PDF corpus based on specific conditions, dynamically loading their content. This enables you to seamlessly insert relevant data from these PDFs into your chatbot's context.