Where knowledge management meets AI: Solutions, approaches, and considerations

While data garners most of the attention when discussing ways to enable successful AI, knowledge plays a pivotal role in how that AI understands and interprets data successfully. Between effective knowledge capturing, organization, accessibility, and KM best practices, there are a myriad of ways to tame KM for driving accurate, relevant AI outputs.

KM experts joined KMWorld’s latest webinar, Enabling Knowledge-Based AI, to examine key components and best practices for adopting AI-enabled approaches that evolve, extend, and power knowledge systems.

John Chmaj, senior director, KM strategy at Verint, explained that initiating knowledge management is a valuable—but specific—process. The effort of building a knowledge base, connecting assets, creating a great search experience, and leveraging those answers across the business is easier said than done. Through bots, Chmaj argued, KM can become an accessible, effective component of any enterprise, quickly.

The Verint Knowledge Automation Bot, for example, allows organizations to benefit from the right knowledge being fed into AI agents without having to rebuild or re-write all their content.

The Verint Knowledge Automation Bot unifies search across all knowledge sources, offering search results summarized by GenAI underpinned by high-quality answers. When content curation is required, information can be fed into the back-end of the bot, jump starting and accelerating migration and curation.

To help accelerate relevance in every search experience with AI, Brian Land, VP, sales engineering at Lucidworks, explained that Lucidworks’ platform combines an open source ethos, LLM-agnostic approach, neural hybrid search, and expert services to drive relevant, accurate AI.

Lucidworks is powered by the goal to shorten the searcher’s journey to find information, ultimately increasing employee productivity, improving customer service, increasing the utility of existing proprietary knowledge, and more. Land noted that GenAI accelerates the progress to this goal, offering:

• Answers to a question rather than a set of documents that may have the answer
• Effective inferences of a query
• Support for a conversational mode of investigation
• Data enrichment during data ingest

Land additionally laid out the path to GenAI success, from defining the use case all the way to tracking the business case. In the end, balancing value and risk, protecting access, mitigating costs, ensuring effective measurement, and cultivating a dynamic, ensemble-based experience forwards true knowledge-based AI.

David Seuss, CEO at Northern Light, focused on the importance of competitive intelligence as it relates to knowledge-based AI, emphasizing that conversational AI has a dramatic impact on business strategy work.

Pointing to a Harvard Business School study with 758 consultants at BCG, Seuss detailed how a group of consultants given a series of business strategy research tasks finished tasks 25% faster with 40% higher quality compared to the group that did not. Clearly, the significance that GenAI and conversational AI experiences has on the efficacy and efficiency of business strategy work is exponential.

Yet, for GenAI to work optimally—especially with popular strategies like retrieval-augmented generation (RAG)—a robust, vetted system of knowledge management must exist.

With a solution like Northern Light’s SinglePoint platform, organizations benefit from a custom-built enterprise knowledge management platform that seamlessly integrates and enables full-text search of every research resource. With GenAI summaries and insights paired with citations and links for sourced information, enterprises can leverage that knowledge-based AI that creates better, optimal business research outcomes with ease, according to Seuss.

At the core of this conversation is how to capture value with AI systems, according to Antero Hanhirova, founder and chief innovation officer at Happeo. Many minds go to chatbots—with the prevalence of solutions like ChatGPT, chatbots are a large representation of how the public sees AI value.

Hanhirova challenged this notion, explaining that chabots are not the most efficient interface for users to interact with computer systems. “We shouldn’t restrict ourselves into thinking that we just want text that we can ingest,” Hanhirova explained, where going beyond text can expand the way we capture and understand knowledge and information.

Generating text is just the tip of the AI iceberg, according to Hanhirova, where workflow automations and recommended actions are characteristics of an elevated, knowledge-based AI system. To achieve this, combining ready-made instructions with generative aspects of LLMs—using reasoning to provide those instructions—ultimately enables outputs to create pages or posts, surface knowledge gaps, curate content, alert experts, and more. This is the key toward driving effective user experiences that expand what GenAI can do.

For the full, in-depth discussion of knowledge-based AI, featuring detailed explanations, a Q&A discussion, and more, you can view an archived version of the webinar here.