Knowledge Buddy is Pega’s AI-powered assistant for end users. Unlike Autopilot and Blueprint which improve developer productivity, Knowledge Buddy improves the experience of the people actually using your application.
The idea is simple. Instead of a customer service agent digging through a regulation document every time a question comes up, they ask the buddy. The buddy retrieves the relevant content and summarises an answer – grounded in your own documents, not random internet content.
How It Works – RAG Architecture
Knowledge Buddy is built on the Retrieval Augmented Generation (RAG) pattern. You upload your documents – PDFs, text content – and they get chunked and stored in a vector database. When a user asks a question, Pega makes two API calls. First, it searches the vector database for the most relevant chunks. Second, it sends those chunks to an LLM (GPT) with a system prompt that says “answer only based on the context provided.” The response comes back grounded in your content.
This is why Knowledge Buddy avoids hallucination. The LLM is not making things up from the internet – it is summarising data you fed it.
Setting It Up
Content is organised in a hierarchy – collections contain data sources, data sources contain content. You create a data source, upload your documents and configure access roles for who can manage and use the content.
Then you create a Knowledge Buddy, point it at your data source and configure the prompt. The default system prompt already grounds the model well – instructing it to stay within the provided context.
For Pega Cloud, the vector database is provisioned automatically per customer in a multi-tenant setup. For on-premise, additional setup is required.
Embedding It in Your Application
In the customer service portal, the Knowledge Buddy widget is available via the control panel under Interaction – Knowledge. Set your buddy as the default and enable the feature. It appears as a tab in the interaction portal that agents can use while handling cases.
For traditional Pega Platform applications, you need to install the Knowledge Buddy gadget component from the Pega Marketplace first.
Watch the Full Walkthrough
In the video below I walk through the full setup end to end – creating a data source for EU 261 flight claim regulations, uploading the PDF, creating a Knowledge Buddy, embedding it in the customer service portal and asking live questions to see the RAG-powered responses. I also trace the API calls in Dev Studio to show exactly what happens under the hood.
Knowledge Buddy is genuinely useful for regulated industries where agents constantly reference policy documents. This video shows how straightforward it is to set up.
