Skip to main content
Verslay
All Use Cases
Customer Success

Customer Support Knowledge Base from Resolved Tickets

Turn the questions your team answers over and over into publish-ready help-center articles — every week, automatically.

Best forAgencyB2BSaaSServices
Agents12 required
Duration4-7 minutes

This recipe reads the tickets your team resolved this week, clusters them into recurring question patterns, and distills each repeat topic into a clean help-center article grounded strictly in the real resolving replies — no invented steps or policies. Articles are mapped to the right help-center section, given a deflection estimate from observed volume, and staged for one-click publish to Notion or Confluence. Nothing writes to your help center until you approve, and a persisted ticket ledger means re-runs never re-mine the same resolutions.

How it runs

Multi-agent orchestration — here's the flow, step by step.

01

Pull resolved/closed tickets in the window, capture question + resolving reply + tags, record the stable ticket-ID dedupe key and skip tickets already in the ledger.

support scanner
01

Sequence the phases, hold the in-memory working set, and enforce the halt condition when no support source is connected.

workflow orchestrator
02

Normalize each ticket into a canonical topic, product area, and question-type, treating existing tags as a prior only.

ticket classifier
02

Score customer language per topic to surface high-friction repeat questions for prioritization.

sentiment analyst
02

Cluster labeled tickets, count frequency per topic, apply the recurrence floor, and produce the ranked repeat-question table.

data analyst
03

Read the resolving replies across each topic's tickets and distill the single correct answer, flagging conflicting resolutions.

knowledge synthesizer
03

Shape short high-frequency topics into concise grounded Q&A entries.

faq generator
03

Expand richer troubleshooting/how-to topics into full articles with steps, summary, and escalation footer, carrying a provenance block.

knowledge base writer
03

Enforce brand voice, terminology, and reading level across every drafted article.

brand voice agent
04

Map each article to its correct Notion/Confluence section and target new pages or existing-page updates.

cms publisher
04

Compute the per-article deflection estimate from observed ticket volume and assemble the run report.

report formatter
04

Stage the CMS payloads, optional Slack digest (after the suppression gate), and the ledger update — nothing executes until the user approves.

distributor

Required Agents

12
  • support-scanner
  • workflow-orchestrator
  • ticket-classifier
  • sentiment-analyst
  • data-analyst
  • knowledge-synthesizer
  • faq-generator
  • knowledge-base-writer
  • brand-voice-agent
  • cms-publisher
  • report-formatter
  • distributor

Connections

Required

notionslackzendesk

Optional

notionconfluenceslackgoogle_sheets

What it does

  • Pulls only resolved/closed tickets in a weekly window from Zendesk or a sheet-of-record
  • Clusters tickets into canonical topics and detects genuinely repeated questions
  • Enforces a recurrence floor so only real demand becomes an article
  • Drafts FAQ and how-to articles grounded in the actual resolving replies
  • Every article carries a provenance block of source ticket IDs + resolving agents
  • Flags conflicting resolutions and SME-confirmation gaps instead of guessing
  • Maps each article to the right Notion/Confluence section and stages for one-click publish
  • Deflection estimate per article from observed ticket volume; idempotent via a persisted ticket ledger

Example prompt

Every Monday, go through the support tickets we closed last week, find the questions that keep coming up, and draft help-center articles for them so we can publish to our Notion help center and stop answering the same thing over and over.

Ready to deploy Customer Support Knowledge Base from Resolved Tickets?

Start free. One click, full agent orchestration.

Get Started Free →