Knowledge Base Analysis & Authoring

Knowledge base analysis looks at your resolved ticket history to find topics that come up repeatedly but are not covered by your existing FAQ articles. It then helps you generate draft articles and publish them directly to osTicket — turning your team’s collective experience into self-service resources.

What Does It Do?

Over time, your helpdesk accumulates a wealth of institutional knowledge in resolved tickets. Knowledge base analysis surfaces that knowledge by:

• Identifying gaps — finding ticket topics that are not covered by any existing FAQ article.
• Clustering similar tickets — grouping related resolved tickets together so you can see patterns at a glance.
• Generating draft articles — using AI to write FAQ-format articles based on clusters of similar tickets.
• Publishing to osTicket — pushing approved drafts directly into your osTicket knowledge base as new FAQ articles.

The goal is to help your team deflect future tickets by giving end users answers before they need to open a support request.

Where Does It Appear?

Knowledge base analysis is available in the Flexivity AI admin console under Analytics > KB Insights. It is not accessed from within osTicket directly — administrators manage it from the admin console.

Note: This feature is designed for administrators and team leads who manage the knowledge base. Individual agents do not need to interact with it directly, though they benefit from the articles it produces.

How Does It Work?

The analysis follows a pipeline that turns raw ticket data into actionable knowledge base improvements.

Gap Analysis

The system clusters your resolved tickets by topic, grouping together tickets that describe similar issues. It then compares those clusters against your published FAQ articles to determine which topics are already covered and which are not.

• Covered topics — a cluster of similar tickets has a matching FAQ article that addresses the same issue.
• Gap topics — a cluster of similar tickets has no corresponding FAQ article. These are candidates for new articles.

Coverage Detection

Coverage detection tells you how much of your ticket volume is addressed by your existing knowledge base. If 70% of your ticket clusters have matching FAQ articles, your coverage is 70%. The remaining 30% represents opportunities to create new self-service content.

Draft Generation

For any gap topic, you can ask the AI to generate a draft FAQ article. The draft is written in a question-and-answer format based on the information from the ticket cluster, including:

• A clear title framed as a question end users would ask
• A step-by-step answer drawn from how your team resolved similar tickets
• Relevant details and caveats mentioned across the cluster

Drafts are starting points. You should review and edit them before publishing to make sure they match your team’s voice and are accurate for your environment.

Push to osTicket

Once you are satisfied with a draft, you can publish it to your osTicket knowledge base with one click. The article is created as a new FAQ entry in the topic and category you select.

The Recommended Workflow

For the best results, follow this workflow:

1. Sync KB articles — Make sure your existing FAQ articles are synced so the system has an accurate picture of what is already covered.
2. Run analysis — Start a new analysis run from the KB Insights page. The system will cluster your recent resolved tickets and compare them against your synced FAQ articles.
3. Review candidates — Browse the gap topics the analysis identified. Focus on clusters with the highest ticket counts — those represent the topics where a new article would have the most impact.
4. Generate drafts — For the gap topics you want to address, generate AI-written draft articles.
5. Review and edit — Read each draft carefully. Edit for accuracy, tone, and completeness. Add any details the AI may have missed.
6. Push to osTicket — Publish the finalized articles to your osTicket knowledge base.

Tip: Run analysis on a regular cadence — monthly works well for most teams. This ensures you catch new patterns as your ticket mix evolves over time.

Configuration

Administrators can adjust two key settings in the KB Insights section:

• Coverage threshold — The minimum similarity score required for a FAQ article to be considered a match for a ticket cluster. Lowering this threshold means more clusters will be marked as “covered,” while raising it surfaces more gap candidates.
• Minimum cluster size — The minimum number of similar tickets required to form a cluster. A higher value focuses the analysis on frequently recurring topics, while a lower value catches niche issues that may still warrant an article.

Tips for Effective Knowledge Base Building

• Sync your KB articles before running analysis. If your FAQ articles are out of date in the system, the gap analysis may flag topics that are already covered. A fresh sync ensures accurate coverage detection.
• Focus on high-volume gaps first. Clusters with more tickets represent the topics that generate the most support load. Addressing those first gives you the biggest return on effort.
• Edit drafts before publishing. AI-generated drafts are a strong starting point, but they may need adjustments for accuracy, tone, or organization-specific details. A quick review ensures quality.
• Track deflection over time. After publishing new articles, monitor whether ticket volume decreases for those topics. This validates that your knowledge base improvements are working.

Troubleshooting

The analysis shows no gap topics. This usually means your existing FAQ articles already cover the main ticket patterns, or the coverage threshold is set too low. Try raising the coverage threshold to surface more candidates.

A gap topic does not seem relevant. Not every cluster warrants a new article. Some clusters may represent one-off issues or internal processes that do not belong in a public-facing FAQ. Use your judgment and skip clusters that are not a good fit.

The generated draft is missing details. Drafts are based on the information available in the ticket cluster. If the tickets in the cluster lacked detail, the draft will too. Edit the draft to add the missing information based on your team’s knowledge.