Strategies for providing accurate answers at scale with Guru

Enterprises struggle to keep knowledge consistent and easily accessible. Bridging the knowledge gaps throughout the entire organization is challenging but can be achieved through modern automation techniques, scalable knowledge delivery, strategic content curation, intelligent routing, and attention to updating information in knowledge hubs.

KMWorld recently held a webinar, Closing the Knowledge Gap: Strategies to Deliver Accurate Answers at Scale, with Rick Nucci, CEO of Guru, who discussed ways to bridge this gap.

According to Nucci, inaccuracy is the most cited risk in production, above security and compliance, when it comes to trusting information. Your knowledge now has two audiences: people and agents.

But AI agents can amplify errors by 1000% from one wrong answer, Nucci explained. The increased speed of change leads to knowledge lags. When the official source can't be trusted, people build their own… and now agents are building them too.

“This is what we call drift,” said Nucci. Drift is when your team operates in one reality and your knowledge describes another. Humans move on and tell each other. Your AI is still reading what got written down last quarter and acting on it with total confidence.

According to Nucci, the loop that closes the knowledge gap consists of the following capabilities:

• Signals arrive: Chats, meetings, tickets, and docs. Wherever answers get given.
• The gap gets found: Contradictions, missing answers, and content that has drifted.
• A draft gets written: Changes arrive as drafts. Nothing publishes silently.
• Your expert approves: Your owners publish. Nobody is bypassed.
• Verified, published, dated: And that becomes the signal that starts the next cycle.

Maintenance stops being a project. It becomes a cycle that runs whether or not anyone remembers to run it.

Nucci recommended four things teams can do today:

1. Take your top 20 articles by traffic. Give each one a named owner and a review date.
2. Find the channel where your team actually answers questions. Start capturing info from it.
3. Set an expiration policy. Content with no verified date is untrusted by default.
4. Ask whoever runs your AI pilot which content it reads, then check whether that set is verified.

For the full webinar, featuring a more in-depth discussion, Q&A, and more, you can view an archived version of the webinar here.