Connect Front and BeforeQuery indexes resolved conversations from your shared inboxes into your knowledge base. Solutions your team wrote once become reusable, citable answers across the widget, bots, and support-form deflection.

Prerequisites

  • A Front API token with read access to the inboxes you care about — create one in Front under Settings → Developers → API tokens.
  • A BeforeQuery workspace owner role.

Data ingested

One resolved conversation, one document:
  • The subject as the title
  • The whole thread, each message labelled Customer or Support Agent
A conversation already indexed is never re-fetched.
A conversation has no public address, so citations to one render as plain text rather than a link.

Set up conversation indexing

Front indexing runs through a workspace-level helpdesk integration plus a knowledge base Resolved Tickets source:
1

Connect Front to your workspace

In the dashboard, open Workspace → Helpdesk Integrations, add Front, paste the API token, and link the integration to your knowledge base.
This same connection powers copilot drafts and deflection for your Front queue.
2

Add a Resolved Tickets source

Open Knowledge Base → Sources → Add source, choose Resolved Tickets, select the Front integration, and click Add. Resolved conversation threads are normalized and indexed like any other document.
3

Verify

Once the sync finishes, check Documents for conversation threads and test a recently solved question in the Playground.
Customer conversations contain PII. Enable PII masking so emails, phone numbers, and identifiers are masked before storage and before any LLM call.

Keeping content in sync

The Resolved Tickets source re-syncs every 5 minutes — there is no schedule to configure, and Sync on the source’s page runs one now if you cannot wait. Only changed content is re-embedded. See Sources Overview.

Troubleshooting

  • Nothing indexed? Check the token can read the relevant inboxes and that conversations are archived/resolved.
  • Keeping conversation knowledge internal? Mark those documents restricted so they answer only for workspace members, or isolate them in an internal source group.