Knowledge gaps queue
An editorial backlog driven by real ticket and search demand.
A two-table signal model (canonical entity plus polymorphic signals) lets manual flags, ticket clusters, and failed searches all land in one queue. Resolving a gap cascades a "resolves" link to every ticket attached to its signals. Phase 2a–2c is shipping today.
Two-table signal model
Shippingknowledge_gaps is the canonical entity (open → drafting → resolved → dismissed). knowledge_gap_signals carries raw evidence with a polymorphic source key and signal-type-specific JSON payload.
Manual flagging
ShippingAgents flag a ticket from the sidebar + menu. Flagged tickets show an amber "Flagged for documentation" pill linking to the queue. The action is idempotent: re-flagging an open ticket attaches a fresh signal to the existing gap.
Cluster detection
ShippingAuto-detector groups closed-without-resolves-link tickets by category, device model, and channel. Confidence scales with cluster size; payload carries ticket IDs, sample titles, and facets.
Failed-search detection
ShippingDoc-scoped searches log to search_query_log with no per-user attribution. The detector aggregates zero-result queries with two or more recurrences.
Editorial workflow
ShippingTwo-pane responsive layout at /documentation/gaps. One-click resolve cascades "resolves" links to every signal-attached ticket; dismiss for non-actionable gaps.
Coverage metric
Shippingimpact_score is a count of unique tickets the gap covers plus search occurrences. Concrete demand, not an abstract weighted score. The badge label adapts to whichever signal dominates ("5 tickets" vs "12 searches").
AI gap detection
PlannedPhase 3 of the docs/KB redesign. The schema is already set up so suggested_title, suggested_outline, and suggested_content can be added without a breaking migration.
Want to see it in action?
Join the waitlist for early access, or browse the rest of the catalogue.