Every incident. Ever.

InvestigatingCustomers in the EU region are experiencing delayed webhook deliveries. Acknowledging the issue and looking into it.

IdentifiedRoot cause is a Frankfurt-region message queue backup. Capacity increase rolling out now.

FixingBackup is processing. Webhook delivery latency dropping. Expecting full recovery within 10 minutes.

ResolvedAll webhook deliveries are processing normally. Total impact: 46 minutes of degraded performance, 0 dropped events.

46 minutes · 0 dropped events · backlog processed in order · Read the postmortem ›

InvestigatingA subset of inbound voice calls in the Mumbai region are not connecting. Investigating with our Twilio routing partner.

ResolvedIssue was on Twilio's side and has been resolved. Total impact: 7 minutes. Affected calls were automatically retried.

7 minutes · Affected calls auto-retried; no manual action needed

IdentifiedA database query optimization regression is causing slow page loads on the conversation history view. Rolling back to the previous deploy.

ResolvedRollback complete. Dashboard performance restored. Service was always available — just slower than usual.

16 minutes · Service available throughout — just slower than usual

MonitoringScheduled Frankfurt RDS failover starting. Traffic shifting to Virginia.

ResolvedFailover complete. Frankfurt back online. No customer impact observed.

23 minutes · Planned. Voice routing failed over to Virginia automatically.

InvestigatingReports of SMS delivery delays specific to Vodafone Idea (India). Twilio confirms upstream issue.

IdentifiedVodafone India congestion. Automatic failover to alternate route enabled for affected customers.

ResolvedVodafone routing restored. All queued messages delivered. Affected window: 1h 4m, ~12% of Vodafone Idea numbers.

1h 4m · ~12% of Vodafone Idea numbers saw >2 min delivery delay

InvestigatingA small percentage of API requests are returning 504. Investigating load balancer health.

ResolvedUnhealthy node removed from rotation. Traffic stable. 0.3% of requests affected for 11 minutes.

11 minutes · ~0.3% of API requests returned 504

InvestigatingVoice conversations timing out or failing to generate responses. Investigating immediately.

IdentifiedPrimary model provider experiencing widespread degradation. Failing over to secondary provider.

FixingSecondary provider absorbing load. Voice service restored for ~70% of customers. Working on remaining traffic.

ResolvedAll voice traffic restored. Final impact: 52 minutes, peak 37% of voice conversations affected. Postmortem to follow within 5 business days.

52 minutes · Voice unavailable for ~37% of conversations. Auto-failover restored service. · Read the postmortem ›

InvestigatingReports of slow web widget loads. Cloudflare global degradation suspected.

IdentifiedConfirmed Cloudflare EU edge degradation. Widget JS still serving but with added latency.

ResolvedCloudflare restored. Widget loads back to normal. No conversation impact during the window.

38 minutes · Widget JS load delayed 200–800ms in EU. New conversations unaffected.

MonitoringScheduled Postgres major-version upgrade begins. Dashboard read-only for the duration.

ResolvedUpgrade complete. Dashboard writeable. No data lost. Conversation traffic was never impacted.

1h 17m · Dashboard read-only. Voice/SMS/Web unaffected.

InvestigatingCustomer reports of webhook signature verification failures after our scheduled key rotation.

IdentifiedOur rotation announcement missed a subset of customers. Rolling back to dual-sign mode immediately.

ResolvedDual-sign mode re-enabled. All webhooks accepted. Rotation will be re-attempted with longer notice. Postmortem to follow.

28 minutes · ~4% of customer webhook endpoints rejected payloads with old key · Read the postmortem ›

InvestigatingAirtel short codes failing on delivery. Twilio and DLT registry investigating.

IdentifiedDLT routing table update propagation delay confirmed by Airtel. Auto-failover to long codes for affected customers.

ResolvedAirtel short codes restored. Affected messages re-routed during the window — 0 dropped, ~2.4% delivered via long code instead.

2h 4m · Airtel short codes unable to deliver. Customers fell back to long code automatically.

IdentifiedStripe webhook ingestion queue backed up. Charges processed normally upstream; only our display lagged.

ResolvedQueue drained. Billing events caught up. Customer-facing charges were never delayed.

19 minutes · Billing events showed up in dashboard ~20 min late. No charges affected.

MonitoringScheduled voice model upgrade rollout begins — 25% canary.

InvestigatingCanary group showing 2× latency. Pausing rollout, investigating.

ResolvedCanary rolled back. Latency back to baseline. Postmortem published — caused by an unintended tokenizer change in the new build.

1h 38m · Voice latency 2× normal for ~22% of conversations until rollback · Read the postmortem ›

IdentifiedIndia DLT registry is down for scheduled maintenance. New SMS template submissions queued.

ResolvedDLT registry back online. Queue drained. All template submissions processed.

5h 0m · New SMS template submissions queued during DLT registry outage. Existing templates unaffected.