Product
Measure it. Prove it. Tell the customers it hit.
Observer reads the metrics you already collect, keeps a measured record of every incident and tells the customers it affected. Your AI SRE finds the cause. Observer proves the impact.
Measure
Status from your own metrics.
An agent inside your network reads the numbers your engineers already trust and sends out one value and its status per check.
- SourcesEvery source Observer reads, and how each one becomes a status.
- Agent & securityThe agent runs in your network and only dials out. Credentials stay on your host.
- MetricsPrometheus, OpenTelemetry, CloudWatch and network probes.
- LogsA Loki or Elasticsearch count, read as a metric.
- DatabasesRead-only queries on Postgres, MySQL, Redis and MongoDB.
- HeartbeatsCron jobs, backups and workers that ping when they run.
- The agentOpen source, inside your network, outbound only.
- Set up with your coding agentPaste one prompt; your coding agent wires up the rest.
Prove
A record your customers can verify.
The same measurements become SLA attainment per customer, signed statements and the numbers leadership reads.
- SLA reportingPer-customer targets, monthly statements and the credit owed, signed.
- Verify a statementAnyone holding a statement can check it against our published keys.
- Reliability CenterPublic commitments and attainment on your status page's own address.
- SLO monitoringError budgets read from the same metrics as the page.
- Executive summaryAttainment, risk and time to recover, for leadership.
Tell
The right customers, in time.
Status pages for every audience, incidents scoped to the customers they hit, and the channels your team already watches.
- Status pagesPublic pages on your domain, in your theme.
- Incidents & AI updatesDrafted when a breach holds, published by a person.
- AI assistCustomer updates drafted from the measured evidence, never the cause.
- Customer pagesEach customer sees their own status against their own target.
- Internal teamsPrivate pages behind sign-in, a password, an IP allowlist or JWT.
- NotificationsSlack, Teams, PagerDuty, Jira, webhooks and email. Recoveries always deliver.
- IntegrationsAlerts in from your tools; findings from your AI SRE as notes.
Developers
Status as code.
Config lives in your repo and ships from CI. Everything else goes through the REST API or MCP.
YAML configobserver.yaml
CLIPlan, apply, and export what you already have.
GitHub ActionPlans on pull requests, applies on merge.
REST APIEverything under /api/v1, described by OpenAPI.
MCP serverOne tool per API operation, under your key's scopes.
Set up with your AI agentPaste into Claude Code, Cursor or Codex. You approve one link; the agent does the rest.
Set up Observer for me. Instructions: https://use.observer/llms.txt
Upserts by key. Console edits to managed objects show as drift.