Book DemoStart Free
System Status

Everything we run, and how it is doing.

Live service availability, 90-day uptime and a plainly written history of what has gone wrong. We report incidents whether or not you noticed them.

Checking current status…

Fetching live availability from the status service.

99.98%

Uptime, last 90 days

Checking… · Report a problem

Services

Chat widgetOperational
Team InboxOperational
AI AgentOperational
Automation engineOperational
Integrations & webhooksOperational
APIOperational
DashboardOperational

Availability, last 90 days

One bar per day across all services. Amber marks a day with degraded performance on at least one service.

99.98%
90 days ago No incidents DegradedToday

Availability in numbers

The same incident history below, counted two ways: when the downtime happened, and what caused it.

Recorded downtime by month

Minutes of customer-visible degradation in the last six months. A month with no bar had no recorded incident.

51m
Mar
0
Apr
95m
May
0
Jun
45m
Jul
0
Aug
Recorded downtime and resulting uptime by month
MonthDowntime (minutes)Uptime
March 20265199.886%
April 20260100.000%
May 20269599.787%
June 20260100.000%
July 20264599.899%
August 20260100.000%

Which service the downtime came from

Every minute of degradation in the last 12 months, attributed to the service that caused it.

191minutes
  • Team Inbox95m 50%
  • AI Agent51m 27%
  • Integrations & webhooks45m 23%
Downtime by service over the last 12 months
ServiceDowntime (minutes)Share
Team Inbox9550%
AI Agent5127%
Integrations & webhooks4523%

Incident history

Every incident with customer impact, including the ones that were fixed before most people noticed.

Resolved2 July 202614:20 – 15:05 UTC · 45 minutesIntegrations & webhooks

Delayed webhook deliveries

Webhook deliveries queued instead of sending. Chat, the inbox and the API were unaffected, and no events were lost.

Cause. A queue consumer failed to restart after a routine deploy and the backlog was not alerted on until it crossed an hour of lag.

What we changed. Added an alert on consumer lag at five minutes, and a health check that fails the deploy if a consumer does not come back.

Resolved18 May 202608:05 – 09:40 UTC · 1 hour 35 minutesTeam Inbox

Slow inbox loading for some workspaces

Conversation lists took up to 20 seconds to load for larger workspaces. Messages continued to send and arrive normally.

Cause. An index was dropped during a schema migration and not recreated, so conversation queries fell back to a full scan.

What we changed. Migrations now run against a production-sized dataset in staging, and query plans are compared before release.

Resolved9 March 202619:12 – 20:03 UTC · 51 minutesAI Agent

AI Agent responses degraded

AI answers were slow or unavailable. Conversations handed over to human agents automatically, so no customer was left without a reply.

Cause. An upstream model provider incident. Our fallback existed but required a manual switch.

What we changed. Failover to a secondary provider is now automatic, triggered on error rate rather than by hand.

Nothing else with customer impact has been recorded in the last 12 months.

Status updates

Get told when something breaks

We email at the start of an incident, when it is mitigated, and once with the post-mortem. Nothing else goes to this list.

Incident notifications only. Unsubscribe in one click.

Something wrong for you only?

Contact support with your workspace ID and we will look.

Help Center

Installation, settings and troubleshooting guides.

Security

How we protect conversation data and what we commit to.