Deep Dive: Knowledge Center - Gaps & Duplicates

[rob_martoncik]

A Deep Dive look at Knowledge Center's Gaps & Duplicates features

This Platform Academy continued a 3 part series breaking down the features of Knowledge Center that started with Article Optimization last session, and will continue with Health Score in Q2's session.  In this session we'll review how the Knowledge Center's insights recommend common task clusters that may not have Knowledge Article coverage and how generative accelerators and the new editor with open prompt can help close the gaps quickly.  We then take a look at how potential duplicates can cause user confusion or provide outdated information and how the Knowledge Center uncovers these items and helps accelerate the handling of merging and/or retiring those issues.

A Quick Knowledge Center Refresher

We kicked things off with a reminder of why Knowledge Center exists: it's your single pane of glass for keeping your knowledge library healthy, accurate, and ready to power everything from agent workflows to AI search and virtual agent responses.

Since its December launch, Knowledge Center has been on a roll. The Q1 release brought:

• Expanded Advanced Editor capabilities, including Knowledge Block support
• Access to an HTML code editor
• Table blocks as a new content option
• Native article approvals inside Knowledge Center
• A new length optimization scan (with an article length counter!) to flag content that's too short to be indexed well — or too long and confusing
• The headline feature of the day: merge duplicates

Closing the Gaps: Knowledge Gap Identification

Next, we walked through how Knowledge Gaps works its magic: it scans your closed/resolved incidents and cases, spots trending topics that don't have a matching article, and surfaces those clusters right on your Knowledge Center homepage.

Once a gap is identified, you've got four options:

• Report a knowledge gap — creates a feedback task an author can pick up later
• Create a knowledge article — pulls the selected incidents into Now Assist's open prompt so you can draft an article in seconds
• Mark as resolved — once you've acted on a cluster
• Unmark as gap — if it turns out there's no real gap after all

Good to know:

• Enable it via Now Assist Admin > Platform > Knowledge — separate skills for ITSM and CSM
• Default filters pull closed/resolved records from the last 90 days, but as of the March release, filters are fully customizable
• Default run frequency is monthly (adjustable to 3 months, 6 months, or a one-time run)
• Each job run consumes Now Assist units
• Currently scoped to incidents and cases only; HR case support is on the roadmap, along with a feedback loop to ML so dismissed clusters don't keep resurfacing

Tackling Duplicates — and Merging Them

Next up we discussed potential duplicate knowledge content. If two or more articles share roughly 80% content similarity, Knowledge Center flags them — regardless of knowledge base or template. Fewer duplicates means better search relevancy and less guesswork for agents and self-service users alike.

The big Q1 addition here is Merge Into New, which lets you combine duplicate articles into a single, cleaner article instead of manually reconciling them. A few important details:

• Merged articles are always created fresh using a standard template — existing articles can't be edited in place
• Source articles must share the same domain and topic cluster, and can't be outdated or retired
• You can merge up to 5 articles at a time, and merging currently works for English-language content only
• When a merged article publishes, source articles that were already published get retired and replaced (governed by your retire-and-replace system property); drafts or in-review versions simply expire
• Article owners get notified automatically when their content is retired in favor of the new merged version
• Enable both skills separately via Now Assist Admin > Platform > Knowledge: Identify Duplicate Articles and Merge Duplicate Articles (merge requires identification to be enabled first)
• Assist consumption is charged for the identification run and per each requested merge action (similar to a generation)

Rapid-Fire Q&A Highlights

• Is this the same as Predictive Intelligence? No — it's a new capability (using the Knowledge Action Framework under the hood) that you simply enable via Now Assist Admin.
• Does it consider Knowledge Blocks? Not yet, for either gaps or duplicates.
• Can I filter what gets clustered? Yes — for both gaps and duplicates, using field-based filters in Now Assist Admin.
• What about articles that get retired? They'll follow your existing retire-and-replace flow.
• What's on the roadmap? HR case support, a feedback loop so unmarked gaps/duplicates don't keep reappearing, article comparison views, duplicate checks at the point of article creation, and more flexible home page filtering.

What's Next

This wrapped up the Platform Academy knowledge series for the spring — the team will be back on May 26th. In the meantime, catch the crew live at Knowledge, where they'll be running labs and sessions on Knowledge Center. And if you haven't already, take a few minutes to fill out the knowledge feedback survey mentioned during the session — it directly shapes what the team builds next.

Missed the live session? The full recording is available in Community. See you at the next Academy!