Pegasystems unveils AI assistant for knowledge management
The workflow automation vendor aims to simplify informational querying processes with an iterative conversational interface that can automatically search all data sources.
Pegasystems on Wednesday unveiled Pega GenAI Knowledge Buddy to simplify knowledge management and knowledge base querying across enterprise systems.
Knowledge Buddy is the latest offering in the Pega Infinity portfolio, which includes Pega GenAI, Pegasystems' suite of generative AI tools.
With the incorporation of generative AI, Knowledge Buddy lets users not only ask questions and provide responses in a conversational way but also retrieve information from any corner of an organization's knowledge base.
Both an enterprise's employees and its customers can interact with Knowledge Buddy to solicit responses that are audited, relevant and secure, according to the vendor. The responses are also contextualized with attributions to the original source material. Users can also collaborate with Knowledge Buddy to write emails and documents based on their knowledge bases.
Enterprises can also build Buddies for different query purposes, including sales, service, marketing and operations. They can be plugged into the organization's various internal systems and digital channels.
In addition, Knowledge Buddy can access libraries in Pega Knowledge Management, a content curation and organization assistance tool. It enables AI-generated content tagging, suggestions and user feedback.
"This is a smart way to extend and expose Pega Knowledge and other formal knowledge bases," Forrester Research analyst Will McKeon-White said. "This will primarily be relying on existing content libraries, which is great for the auditability and governance focus of this tool."
Pega GenAI Knowledge Buddy can also be used by customers as a resource for online self-service, complete with knowledge articles and a knowledge search function.
Focusing on safety
Pegasystems incorporated various safety measures with Knowledge Buddy to ensure that its output is secure and appropriate. It starts during setup, when administrators craft Knowledge Buddy's desired behaviors by describing and testing prompts as well as establishing rules and blocks to limit its scope of access by barring sensitive information.
To help ensure that only authorized knowledge and information updates are included and accessible to Knowledge Buddy, Pega also installed specific controls for administrators to let only approved users manage the content.
Pairing GenAI with RAG
In addition, Knowledge Buddies include Pega GenAI's retrieval-augmented generation (RAG) architecture by default. RAG construction enables a natural language query experience that also ensures organizational knowledge is the only knowledge base consulted. This method is purported to decrease the likelihood of erroneous output. It's an alternative method to training and using a private model, which could be more costly and take more time.
Pega GenAI Knowledge Buddy, along with competing generative AI assistants such as Microsoft Copilot and Google Duet AI, lets organizations use large language models on their private knowledge bases guarded by enterprise privacy and security controls.
"The RAG pattern is, indeed, becoming the most common way for organizations to 'ground' generative AI responses to specific data," said Jason Wong, a Gartner analyst.
RAG can be used for retrieving both public internet data and data from private knowledge bases, Wong noted.
This is a smart way to extend and expose Pega Knowledge and other formal knowledge bases.
Will McKeon WhiteAnalyst, Forrester Research
Typically, seeking the appropriate knowledge to complete a business task could take employees 20% to 30% of their time, according to Wong. One of the main benefits of generative AI, when it's paired with RAG, is that it can significantly reduce the time needed to locate relevant information.
But configuring knowledge base access controls for the RAG pattern is a considerable undertaking and "can add obstacles in widespread adoption across all the organizational knowledge," Wong said.
In addition, generative AI and knowledge management tools can only be as intelligent as the information they are given. Human inefficiency can still endanger the usefulness of a Knowledge Buddy's output. It will still depend on how accurate and tidy the organization's knowledge stores are, especially across a large organization with various systems for storage.
McKeon-White said that dynamic knowledge, or knowledge that does not have a regular sequence, can be particularly challenging to wrangle.
"One of the biggest areas of competition in GenAI systems is dynamic knowledge, and the ability to pull -- accurately -- from both formal and informal knowledge repositories," McKeon-White said.
Pega GenAI Knowledge Buddy will be available on Pega Cloud in the Pega Infinity 2024 release during the first half of 2024, according to Pegasystems.
Mary Reines is a news writer covering customer experience and unified communications for TechTarget Editorial. Before TechTarget, Reines was arts editor at the Marblehead Reporter.
