Research notes

Conversation Listeners

Opt-in listeners that capture conversations from every channel an organization uses — support, email, team chat, customer messaging, webchat, sales tools, voice — and route them into the signal-extraction pipeline with consent and retention rules attached.

Operating principle

Production AI is not a prompt. It is a system of context, tools, permissions, traces, evals, and feedback loops.

How they are built

Each listener runs with the minimum scope required: read access to specified channels, no message-write capability unless a workflow explicitly needs it, no DMs unless a user has opted in. Captured messages are written to a dedicated table governed by Data Foundations with a configured retention window. PII handling is declared per source.

  • Minimum-scope tokens per channel
  • Explicit retention window per source
  • PII redaction rules declared per listener
  • User-visible off-switch and audit log

Channels we typically support

Support platforms via webhook or API (Zendesk, Front, Intercom, Help Scout, Salesforce Service Cloud); email inboxes via Gmail API, Microsoft Graph, or IMAP; team chat via platform APIs (Slack, Microsoft Teams, Discord); customer messaging via the WhatsApp Business Platform, Telegram Bot API, SMS providers (Twilio, MessageBird), Messenger, and Apple Messages for Business; webchat and in-app widgets via SDK hooks; sales tools via CRM activity logs (HubSpot, Salesforce, Pipedrive); voice transcripts from call platforms (Twilio, Aircall, Five9, Genesys, Dialpad). New channels are added as MCP-compatible listeners; the same governance, retention, and consent rules apply regardless of source.

Related resources