Heads-up in Intercom when a customer hits sustained latency

Watches Honeycomb for sustained p95 latency or error-rate degradation scoped to a single tenant, then sends that customer a transparent Intercom heads-up — turning a quietly-slow experience into a proactive we see it and we're on it message instead of a frustrated bug report.

Use it for performance degradation that affects one account at a time (a heavy data set, a hot query, a noisy-neighbor effect) where the customer feels slowness but has no error to screenshot. Good for high-tier accounts you want to over-communicate with.

1. 1A Honeycomb trigger fires when a per-tenant latency SLO threshold is breached for a sustained interval.
2. 2The flow reads the offending query group and tenant ID from the Honeycomb event.
3. 3A logic step confirms the breach is sustained, not a momentary blip, and checks plan tier in Postgres to decide whether to notify.
4. 4Postgres resolves the tenant's primary contact.
5. 5Intercom sends a transparent degraded-performance notice with an expected timeline.
6. 6A Slack alert flags the account to the reliability channel for tracking.