Wrangling and refining content for generative search with NiCE

As organizations rush to adopt generative AI, many struggle with content quality and disconnected workflows. How do you streamline all the knowledge you have within the organization to deliver relevant answers customers are looking for?

KMWorld recently held a webinar, GenAI Workflow in Action: Elevate Content, Elevate Results, with Shelley Reyes, product engagement and strategy from NiCE, CXone Mpower Expert, who introduced proven GenAI content optimization techniques that help knowledge teams streamline updates, resolve common issues, and improve AI performance at scale. 

“Our AI-powered platforms connect people, systems, and workflows to drive smarter work at scale and truly delivering measurable outcomes,” Reyes said.

Managing the process in one place helps knowledge become reliable, easier to find, and actionable, she explained.

This approach includes:

• Capture: Ingest content
• Cultivate: Content adequacy
• Calibrate: Editor AI – Predefined Prompts
• Calculate: Faithfulness and relevancy

“Remember this framework is AI-agnostic,” Reyes noted. “We want you to focus on the overall workflow itself.” These principles can come to life with Mpower Expert.

Generative search powered by RAG will surface correct responses delivered as rich-context responses or articles that are very precise.

Things to consider with Capture are clear summaries allowing AI to retrieve information for correct generative responses. Mpower Expert offers an HTML importer to ingest content from different platforms without losing the faithfulness of the context.

“You’re setting the foundation in your knowledgebase for AI-optimized knowledge,” Reyes said. “You’re making is discoverable for gen search at the very start.”

Next is the cultivation step, which examines if the content is adequate to what customers and agents are asking for. Additionally asking these questions is key for making sure knowledge is gen search ready: What are customers calling about? What are agents searching for? What are they navigating to? These questions need to be answered completely.

“This process is about verifying your content foundation is very strong,” Reyes said.

After capturing and cultivating the content, it needs to be calibrated. This is where the AI Editor within Mpower Expert can enhance prompts for optimization by maintaining consistent tone and brand voice.

The final step is to calculate, where you measure response effectiveness, consistency, and relevance, she concluded. Faithfulness measures the accuracy of the generated answer against the source context whereas relevancy evaluates how well the generated answer addresses the user’s original question.

“This focuses on adding that credibility and trust,” Reyes said. “Following this AI-agnostic framework builds the culture, ownership, and trust that makes your knowledge a real competitive advantage with AI tools.”

For the full webinar, featuring more detailed explanations, a Q&A, and more, you can view an archived version of the webinar here.