Find your way
Which part of this documentation to read first, depending on what you do with Cortex, and where to look up a specific task.
This documentation has a section for each kind of user, and a section that explains the ideas behind Cortex. Pick the one that matches what you do. If you are not sure, read What is Cortex? and take the 10-minute tour first.
Who can do this
Everyone. The sections are not locked: you can read the Studio and Control guides even without a Studio seat or the Admin role. You only see the matching screens in Cortex itself when your role allows it.
What Cortex is, signing in, the tour, who can do what.
Everyone: ask, open, follow, handle tasks.
Builders: dashboards, apps, workflows, models.
IT: people, approvals, data, AI engine, licence.
The ideas behind it, without step lists.
I use Cortex to get answers and open dashboards and apps
Start with the Hub guide. It covers Home, Ask, finding and opening dashboards, apps and workflows, tasks, pinning values and subscribing to digests and alerts.
Go to Find your way around (Cortex Hub)
I build dashboards, apps, workflows or models
You have a Studio seat. The Studio guide shows how to start something with AI, how to review and publish it, and how to look after your semantic models.
Go to Start something and follow your requests
I administer Cortex
You are an admin of Cortex Control. The Control guide covers users and roles, approvals, policy, budgets, data sources, the AI engine, the licence and the audit log.
Go to Overview, attention list and health
I want to understand how it works
The Concepts section explains the ideas behind Cortex without step lists: how the parts fit, semantic models, certified figures, access and security, how Ask and AI work, the different kinds of experiences, workspaces, and a glossary.
Look up a task
Example: where Sam looks
Sam de Vries at Brightwater Foods is both a builder and an admin, so Sam reads from all three sections. On Monday Sam wants to know why a figure differs between two dashboards, so Sam opens Semantic models to see that both read one definition, and then Model checks to see whether the nightly check flagged anything. On Tuesday a colleague asks for access to a dashboard, so Sam opens Approvals.