Examining the future potential of agentic AI at KMWorld 2025
In today’s hyper-competitive business environment, success hinges not just on what you know, but on how fast and effectively you act. While most organizations are swimming in data and insights, turning that knowledge into timely, strategic action remains a major challenge.
At KMWorld 2025, David Seuss, founder and CEO, Northern Light, held his session, “From Insight to Action: Agentic AI Turning Knowledge Into Strategic Momentum,” discussing how agentic AI represents the next evolution of AI.
Unlike traditional AI models that passively wait for user prompts, Agentic AI acts with autonomy.
It can proactively scan massive volumes of enterprise knowledge, detect emerging patterns, surface strategic opportunities, and even initiate next steps—all within the boundaries and priorities defined by your organization.
“The ‘Big Bang’ didn’t happen millions of years ago, it happened in November 2022 when ChatGPT arrived and set off a firestorm of activity,” Seuss said.
What Google, Microsoft, and OpenAI are talking about, a year later these same technologies become more widespread a year later, he explained. He believes that 2026 will be the year of agentic AI.
Seuss shared and demonstrated practical, real-world use cases that show how agentic AI: moves teams from reactive analysis to proactive execution, streamlines competitive intelligence, market research, and decision workflows, delivers personalized, autonomous insights to the right people at the right time and builds lasting strategic momentum by connecting insights, teams, and goals.
Snehit Cherian, CTO, Global Nexis Solutions, LexisNexis detailed the keys to building a high-trust agentic workflow during his portion of the discussion.
Now-established architecture like RAG, along with new standards such as Model Context Protocol, improve the reliability and accessibility of data that LLMs and agents need, but they're only part of the story.
Cherian shared strategies for building confidence in increasingly AI-driven outputs of the kind that tap LexisNexis's industry-largest collection of news data approved by publishers for Generative AI use.
“We have thousands of publishers that have licensed content with us,” Cherian said.
There are a variety of Agent frameworks, this includes ADK, AutoGen, Crew.AI, LangChain, LangGraph, Semantic Kernal, and Strands. There is also A2A and MCP, he explained.
He offered 9 considerations for building agentic systems. These considerations include:
- Start simple
- RAG and Agentic RAG
- Context engineering
- Tool choice optimization
- Automated metrics, evals, and critics
- Transparency and observability
- Human in the Loop
- Error handling
- Cost and latency versus performance
“Is an agentic solution really necessary for this use case? That should be the first question you ask before starting this,” Cherian said. “If so, make sure you put in goals to focus on.”
KMWorld returned to the J.W. Marriott in Washington D.C. on November 17-20, with pre-conference workshops held on November 17.
KMWorld 2025 is a part of a unique program of five co-located conferences, which also includes Enterprise Search & Discovery, Enterprise AI World, Taxonomy Boot Camp, and Text Analytics Forum.