Skip to main content

Operations and automation

Simplr's operations layer turns signals from monitoring, fraud, feedback, releases, and connected tools into one queue with ownership, SLA state, approvals, and an auditable response.

Start with Operations​

Open Dashboard → Operations and choose a product scope. The overview combines:

  • critical incidents, SLA breaches, affected users, and automation outcomes;
  • RUM error rate and sessions, API success/latency, and recent high-risk checks;
  • active and urgent feedback;
  • runs waiting for human approval; and
  • recent inbox work and connected-system health.

Pin the products you operate most often. Selecting an incident opens an investigation-first workspace with its evidence and linked operational context.

Unified inbox and SLA​

The inbox normalizes incidents and feedback without hiding their source records. Teams can filter, assign, acknowledge, and resolve work while Simplr keeps the source status synchronized. SLA policies can be scoped to source, severity, service, environment, and team; each item records acknowledgment and resolution deadlines plus breach state.

SLA policies require Growth. The base operations overview and inbox require Starter.

Connections and external work​

The shared application and environment catalogue maps every operated application to its type, owners, environments, repository, runbook, telemetry IDs, and deployment policy. Connections can ingest Cloud Monitoring, OpenTelemetry, Sentry, inbound support email, and read-only investigation sources. ClickUp work-management connections can create and synchronize external work items from inbox records according to their configured severity and feedback rules.

Test each connection before enabling a rule. Credentials are stored encrypted and are never returned by read endpoints.

Rules, runbooks, and approvals​

A rule matches an event and starts a versioned runbook. Runs retain their input, step state, credit charge, approvals, result, error, and audit history. Built-in and custom runbooks can mix automated work with customer-agent steps.

Sensitive steps pause for an owner/admin decision. Production deployments, merge operations, database mutations, authentication, billing, and infrastructure actions are approval-gated; secret changes and destructive data deletion are blocked. Non-production auto-deployment must target a mapped environment, configured actor, and input-free GitHub workflow, then pass test verification and telemetry checks.

An automation run consumes 2 credits. Creating services, rules, runbooks, approvals, and viewing audit history does not consume credits.

Self-healing incidents​

On Growth, the self-healing control plane groups internal failures and connected telemetry into incidents. A customer coding agent can claim one incident, request a policy decision for each action, implement and test locally, request GitHub authorization, record the executed operation, and attach verification evidence. Simplr keeps an incident open while evidence is incomplete or telemetry still contradicts recovery.

AI investigation costs 5 credits. Simplr supports explicitly connected OpenAI, Anthropic, and Gemini providers, read-only MongoDB collection allowlists, and Google Cloud Logging with a Logs Viewer service account. Evidence is redacted before model analysis.

Notifications and public status​

The Notification Centre routes billing, incidents, feedback, GitHub approvals, releases, integrations, automation, and service-health events. Each member controls personal email/Telegram subscriptions and severity thresholds. Owners/admins can also add verified organization email recipients for operational alerts.

For customer communication, publish a redacted incident or feedback record to /status/:slug, then add timestamped updates using received, investigating, in_progress, verifying, monitoring, resolved, or closed. Unpublishing disables public access without deleting the internal record.

Connect a coding agent​

Use the AI Coding MCP for the complete operator questionnaire, credential boundaries, scheduled queue workflow, GitHub authorization protocol, and verification tools.