Solutions
For teams that keep telemetry in-house
Production Master is aimed first at platform and SRE teams that cannot send telemetry to a vendor.
Everyone else gets the same evidence model, covered below.
Platform and SRE teams
In your own cluster
For platform and SRE teams at 50–500-engineer companies on Kubernetes, running Prometheus, Loki and Argo CD with Grafana in front, who cannot send telemetry to a vendor.
- Self-host via Helm in your own Kubernetes cluster — available in closed beta.
- Self-hosted, you configure your own model provider and keys, so the provider and its terms are your decision.
- The same cited evidence and approval boundary as the hosted product.
- Prometheus and Kubernetes are read through native, read-only connectors. Logs are read through the Grafana connector today; a direct Loki connector is planned, and Argo CD has no connector yet.
On-call engineers
The timeline, already built
The expensive part of an incident is not the fix — it is establishing what changed, and in what order.
Production Master arrives with that reconstruction cited.
- A cited timeline from the first minute, not after the postmortem.
- The chosen explanation in writing with its exhibits, so you are not re-deriving it at 3am.
- A proposed rollback that waits for you rather than firing on its own.
Multi-service teams
One standard, every service
Investigation quality usually tracks who was on call.
A shared evidence model makes it a property of the system: the same exhibits, hypotheses and approval boundary.
- One evidence model across services, teams and time zones.
- Investigations that stay readable after the people who ran them have moved on.
- Self-host via Helm inside your own infrastructure — available in closed beta.
Engineering leaders
Conclusions you can audit
A conclusion without evidence is a claim you must re-derive.
Every report keeps the exhibits under its conclusion, so a later review can check the reasoning.
- Every claim traceable to the log, metric or diff it came from.
- The conclusion preserved with the exhibit behind every claim.
- Remediation kept behind an explicit human approval gate.
It reads your stack
Paging still pages, dashboards still chart, CI still ships.
Production Master sits in the gap between the alert and the decision, and replaces none of them.
Turns the approved signals into exhibits, tests falsifiable hypotheses against them, and records what each one proves or rules out.
The connectors, by name
What each one does today.
Prometheus and Kubernetes are read through native, read-only connectors.
Logs are read through the Grafana connector; a direct Loki connector is planned.
- GitHubSearches code. Comments on pull requests and opens draft pull requests.Shipped
- GrafanaReads logs, metric ranges and SQL queries. Annotates dashboards.Shipped
- JiraReads tickets. Adds comments and labels, and moves tickets between states.Shipped
- SlackSearches channels and threads. Sends notifications.Shipped
- PagerDutyStarts an investigation from an incident. Pages and escalates. Not read as evidence.Shipped
- DatadogStarts an investigation from an alert. Trigger only: Datadog data is not read as evidence.Shipped
- PrometheusNative connector, read-only. Reads metrics.Shipped
- KubernetesNative connector, read-only. Reads cluster events, pods and nodes within the scope you approve.Shipped
- LokiRead through Grafana today, where your Grafana is backed by it. No run reads Loki directly yet.Planned
- Argo CDNo connector yet.Planned
- OpsgenieNo connector yet.Planned
Bring the incident you argued about
The investigation that was hardest to settle is the one worth running through the evidence model first.
Self-host via Helm — available in closed beta.