99.99%
Monthly uptime target, Scale and Enterprise
System status
The services we run, and which of them the 99.99% monthly uptime target covers. An incident notice goes out by email and to a webhook.
5
Services under the agreement
How status is reported today
A person on the team posts here when an incident is confirmed, and the same notice goes out by email and to a webhook. Measured uptime for each service is not published yet. The monitor that will publish it is being built, and the figures land on this page once it runs.
Services we run
Each service is listed with what it does and with its standing under the service level agreement.
AI inference and outbound webhooks run outside the agreement, because a late summary is a delay rather than an outage.
Service
Service agreement
API and gateway
REST API, WebSocket streams and inbound webhook listeners
Service agreementCovered
Channel ingestion
Workers pulling messages from the 10 live networks and the 5 in beta
Service agreementCovered
Message sending
Accepting an outbound message and handing it to the network
Service agreementCovered
Search index
Full-text search across message text, transcripts and file contents
Service agreementCovered
Apps
Web, macOS, Windows, iOS and Android clients
Service agreementCovered
AI inference
Transcription, translation and thread summaries
Service agreementOutside the agreement
Outbound webhooks
Event delivery to client SDKs, MCP servers and your endpoints
Service agreementOutside the agreement
Uptime history
Measured uptime is not on this page today. The agreement sets the method, and the figures follow the monitor.
The agreement defines downtime minutes for each covered service, and monthly uptime as a share of the minutes in the calendar month, carried to three decimal places. Once the monitor publishes against that definition, each service gets its own figure here and the history stays.
Written incident reports arrive the same way. For a critical incident the agreement commits us to a root cause analysis, published here and left as written.
Service level agreement
The target is 99.99% of a calendar month, counted separately for each covered service. A miss puts a credit on the next invoice, and nobody has to file a ticket to get it. That commitment runs on Scale and Enterprise, and the agreement holds the credit schedule. Starter and Team run on best-effort uptime with no credits.
Incident notices
An outage or a maintenance window goes out by email or to a webhook you choose. Ask support to add your address or your endpoint. For a critical incident the agreement sets the deadline for the first post here and for the written analysis that follows.