Telemetry and observability

Logs, metrics, traces, product analytics, and AI/agent telemetry with supportable operating boundaries.

Telemetry helps teams understand reliability, product behavior, integration health, queue lag, realtime delivery, and AI/agent quality. Assistance treats telemetry as an operated system: signals have owners, retention rules, alert routes, dashboards, and support workflows.

What Assistance operates#

Within scope, Assistance can operate metrics collection, log and trace pipelines, dashboards, alert rules, product analytics schemas, AI traces, cost attribution, incident triage, and observability handoff runbooks.

What the customer owns#

The customer owns business event definitions, privacy/consent decisions, sensitive-data classification, who may access raw logs or traces, and product decisions made from analytics.

Core guides#

• Product analytics and telemetry governance
• Managed Prometheus
• Agent Observability
• Production AI governance

Safe-use requirements#

• Do not log secrets, payment data, raw customer content, or prompts unless explicitly approved.
• Use correlation IDs across APIs, queues, integrations, realtime channels, and AI workflow runs.
• Separate raw telemetry access from aggregate dashboards.
• Define retention for logs, traces, product events, and AI prompts/completions.
• Alert on user-impacting symptoms, not only infrastructure saturation.

Support workflow#

Telemetry incidents should include the service or workflow name, environment, time window, dashboard or trace link, expected behavior, actual behavior, and business impact. Planned changes should include dashboard updates, alert review, and rollback criteria.