Triage documentation

Triage turns a support ticket into an answer. A ticket arrives, Triage extracts who it is about, looks up that user's funnel drop-off, finds similar past tickets, and then an AI agent investigates read-only across your own infrastructure — databases, logs, metrics, error tracking, Kubernetes, cloud — and writes up a root cause with the evidence it found.

Nothing is hosted for you. The agent runs on your machines and talks to your systems with credentials you supply.

Two pieces, one product

The portal is the source of truth for configuration. The desktop app pulls that configuration down on a schedule and executes against it. See How configuration reaches the desktop for the mechanics.

Start here

You are setting up an organization for the first time. Read Architecture, then work through the Admin setup checklist. Budget about half an hour: most of it is pasting credentials.

You were sent a login and told to install the app. Go straight to Install the desktop app, then Sign in. You do not need anything from the admin section.

You want to write a good playbook. Anatomy of a playbook explains each field and what the agent does with it; Example playbooks are five complete, copyable ones.

You are evaluating Triage. Roles, plans and seats covers what the free tier gives you and where the paid line falls. Security model covers what leaves your network (very little) and what does not.

Conventions in these docs

Anything shown as code is a literal value: a field name, a key, an environment variable, a URL path. Anything shown in bold is a label you will see on screen.