Cortex is not a chatbot bolted onto a product. It is the layer of Brainis that reads across every module - deals, tasks, people, invoices, candidates - and turns that into answers, analyses, and actions.
What Cortex does
- ›Answers questions grounded in your data. Ask "which deals slipped this month and why" and the answer cites your pipeline, not the internet.
- ›Runs plays. A library of 420+ pre-built business plays: forecasts, diagnostics, prioritizations, and playbooks for every department and role. See Plays.
- ›Produces deliverables. Real documents with citations - briefs, plans, analyses - that you can review, version, and share. See Deliverables.
- ›Works on a schedule. Autopilot runs selected plays automatically and reports back. See Autopilot.
The two surfaces
The Cortex station (the Cortex tab in Mission Control) is chat-first: a conversation with quick access to Plays, Agents, Fleet, and Memory. It is where most day-to-day interaction happens.
Deep Cortex is the full module for people who operate on top of it: dashboards, predictions, scenario planning, a war room, experiments, an ROI ledger, and diagnostic tooling. Reach it from the station.
What it costs
A quick answer costs about 2 credits, a deep analysis about 10. Costs are flat and shown up front; see How credits work.
What it will not do
Cortex proposes; the action gateway disposes. Any change Cortex wants to make to your data flows through the same autonomy, approval, and audit machinery as every agent. You decide how much it may do alone. See Autonomy levels.
Tip: Cortex improves with context. The single highest-leverage thing you can do is complete your company profile and knowledge base - see Memory and context.