34Customer Service

Knowledge Base Auto-Updater

Mines the 3-5 questions your customers asked most this month with no KB article, drafts them, routes for SME review, and publishes.

The problem

Most help centers go stale within 90 days of launch. Reps answer the same questions in 1:1 tickets, the answers never make it back into the KB, search returns "no results found" for the exact phrase customers use, and ticket deflection rate stalls. A KB auto-updater clusters resolved tickets into themes, surfaces the highest-volume questions with no matching article (or with an outdated one), drafts the article in your house style with citations to the resolving ticket threads, routes to the right SME for review in Slack, and publishes to Document360 / Guru / Confluence / Notion on approval.

Typical leak: 12-22% of inbound tickets are repeats of unsearched questions; a 5pp lift in KB hit rate typically pulls 8-15% of ticket volume off the queue

Ticket deflection rate + KB hit rate

8-15% of ticket volume deflected on a 5pp KB hit-rate lift; new-article time-to-publish 3-6 weeks → 2-4 days

Gartner Customer Service Self-Service Index 2024; Intercom State of Customer Service 2024

Integrates with

ZendeskIntercomFreshdeskHelp ScoutHubSpot ServiceDocument360GuruConfluenceNotionSlackClauden8n

How it works

Agent · Knowledge Base Auto-Updater

Cluster scan · 47 tickets / 7d

Gap found · "bulk import" × 11

Claude drafts article + citations

SME review in Slack · Maya

Published to Document360