Knowledge Base Analysis & Authoring

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

What Does It Do?

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

Identifying gaps — finding ticket topics that are not covered by any existing Help Center article.
Clustering similar tickets — grouping related solved tickets together so you can see patterns at a glance.
Generating draft articles — using AI to write article drafts based on clusters of similar tickets.
Writing drafts to Zendesk Help Center — pushing approved drafts directly into your Zendesk Help Center as draft articles, never as published articles.

The goal is to help your team deflect future tickets by giving end users answers before they need to open a support request — using Zendesk Help Center’s built-in semantic search to surface those answers automatically.

Drafts Are Never Auto-Published

This is the most important guarantee of the workflow: Flexivity never publishes Help Center articles automatically. Drafts always land in Zendesk’s draft state and require explicit human review and publish.

The publish step happens entirely inside Zendesk:

Flexivity writes the draft article via the Zendesk Help Center API.
The draft appears in your Help Center under the configured section, with status Draft.
A Zendesk Help Center publisher reviews the draft in Zendesk’s built-in article editor, makes any edits, and clicks Publish when satisfied.

This means Flexivity respects your existing Help Center publishing permissions, review workflows, locale management, and any approval gates you have set up in Zendesk. Nothing about Help Center governance changes.

Note: The Zendesk user that Flexivity authenticates as during OAuth connect must have permission to create draft articles in the target Help Center section. See Feature Configuration — Zendesk for the exact permission requirements.

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 Zendesk directly — administrators manage analysis runs and review draft proposals from the Flexivity admin console.

Note: This feature is designed for administrators and Help Center managers who curate the knowledge base. Individual agents do not need to interact with KB Analysis directly, though they benefit from the articles it produces showing up in Help Center search.

How Does It Work?

The analysis follows a pipeline that turns raw ticket data into actionable Help Center improvements.

Gap Analysis

The system clusters your solved tickets by topic, grouping together tickets that describe similar issues. It then compares those clusters against your published Help Center articles (which Flexivity reads via the Zendesk Help Center API) to label each cluster as covered or as a gap:

Covered topics — a cluster of similar tickets has a matching Help Center article that addresses the same issue.
Gap topics — a cluster of similar tickets has no corresponding Help Center article.

These labels are informational — they tell you whether a matching Help Center article already exists. They do not restrict which clusters you can act on.

Coverage Detection

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

Draft Generation

For any candidate topic the analysis surfaces — covered or gap — you can ask the AI to generate a draft article. Gap topics are the most common starting point because they represent uncovered content, but you can also generate a draft for a covered topic, for example to refresh an outdated article or to author a sibling article for a different angle on the same issue.

The draft is written 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 in Zendesk’s article editor before publishing to make sure they match your team’s voice and are accurate for your environment.

Push to Zendesk Help Center

Once you are satisfied with a draft in the Flexivity admin console preview, you click Push to Help Center. Flexivity creates the article in the section you select using the Zendesk Help Center API. The article lands as a Draft in Zendesk.

Important: Flexivity sets the article state to “Draft” via the Help Center API. Publishing it (and selecting a final locale, section, labels, and visibility) happens inside Zendesk’s article editor by a Help Center publisher.

The Recommended Workflow

For the best results, follow this workflow:

Make sure Help Center is up to date. Flexivity reads your published Help Center articles to determine coverage; out-of-date or unpublished articles can make the gap analysis less accurate.
Run an analysis. Start a new analysis run from the Flexivity KB Insights page. The system will cluster your recent solved Zendesk tickets and compare them against your Help Center articles.
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.
Generate drafts. For the gap topics you want to address, generate AI-written draft articles.
Review the preview. Read each draft in the Flexivity preview. Make any high-level decisions about whether to keep, skip, or regenerate.
Push to Help Center. Click Push to Help Center. The draft lands in Zendesk under the configured section, in Draft state.
Edit and publish in Zendesk. A Help Center publisher opens the draft in Zendesk’s article editor, makes detailed edits, sets the locale and visibility, and clicks Publish.

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 the following settings in the KB Insights section of the Flexivity admin console:

Article format — Choose between Q&A (short question-and-answer style, like a FAQ entry) and Documentation article (longer-form prose with sections and steps). This determines the style and structure of every generated draft on this instance, so pick the format that matches how your Help Center is organized today.
Coverage threshold — The minimum similarity score required for a Help Center 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.

The default Help Center section that drafts land in is also configurable per instance — see Feature Configuration — Zendesk.

Tips for Effective Knowledge Base Building

Keep your Help Center catalog tidy. Archive or unpublish out-of-date articles in Zendesk so gap analysis does not incorrectly flag topics as covered by stale content.
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 in Zendesk before publishing. AI-generated drafts are a strong starting point, but they may need adjustments for accuracy, tone, or organization-specific details. Edit in Zendesk’s article editor where your locale, labels, and section management live.
Track deflection over time. After publishing new articles, monitor whether ticket volume decreases for those topics in Zendesk’s reporting. This validates that your knowledge base improvements are working.

The Flexivity AI integration for osTicket includes a customer-facing AI search widget on the public-facing osTicket pages. The Zendesk integration does not include an equivalent widget for a deliberate reason: Zendesk Help Center already includes built-in semantic search across all Zendesk Suite tiers.

A separate Flexivity widget would duplicate functionality your team is already paying for, and would create two competing search bars on your Help Center pages. Instead, Flexivity focuses on the upstream problem — finding gaps in your Help Center content so your existing Zendesk Help Center search can answer more questions — rather than adding a parallel search surface.

Troubleshooting

The analysis shows no gap topics. This usually means your existing Help Center 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 Help Center. 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 in Zendesk’s article editor to add the missing information.

The push to Help Center failed with a permission error. The Zendesk user Flexivity authenticated as during OAuth connect must have permission to create draft articles in the target Help Center section. See Connecting Zendesk and Troubleshooting for help.

What Your Administrator Can Configure

The KB Analysis prompt, the on/off toggle, the article format (Q&A vs documentation article), the coverage threshold, the minimum cluster size, the default Help Center section, and the writeback enable toggle are all configurable — see Feature Configuration — Zendesk. For what data Flexivity reads from your Zendesk tickets and Help Center, see Data Disclosures — Zendesk.