Skip to content
Wavertex
Documentation menu

First steps

Verified with Cortex on October 2, 2026

This page shows what a person using Cortex every day sees: signing in, chatting with a model and what happens to the sensitive values they type.

Sign in

Open the Cortex address your organization gave you. You have two choices:

  • “Sign in with SSO” sends you to your organization's identity provider;
  • otherwise, enter your “Email address” and “Password”, then click “Sign in”.

If your account requires two-factor authentication, Cortex asks for a 6-digit code. The first time, it first shows a QR code to scan with your authenticator app, then you enter the code it displays. After too many attempts, the page asks you to try again in a few minutes.

Once signed in, you land on the chat, unless your role opens a console first (administrator, auditor, knowledge manager). Your session expires after a delay set by your administrator: Cortex then asks you to sign in again.

Chat

  1. At the top center of the screen, choose a destination and a model. These are the ones your administrator allows for your role. A label shows whether the destination is local or external.
  2. Write your message in the input box, whose placeholder names the destination (“Write to …”). Under the box, the DLP profile that applies is shown.
  3. Send with the Enter key (Shift + Enter for a new line), or with the “Send” button.
  4. The reply appears as it is generated. The “Stop” button interrupts the generation.

Under a reply, you can copy it; under the last one, you can also regenerate it. If the administrator enabled web search or the knowledge base, controls appear in the input box and the sources used are shown under the reply.

Cortex screenshot: a conversation where the message contains a name, an email and a phone number, with the sensitive-items chip, and a reply that repeats those values in clear
An exchange: sensitive values are flagged on your message, and the reply gives them back to you.

Understand redaction

Before sending, Cortex detects the sensitive items in your message (for example emails, numbers, secrets) and replaces them with tokens such as [EMAIL_1]. Your message shows a “× sensitive items” chip: hover it to see the detected types, never a value. Click it to read the full explanation.

The model's provider only receives those tokens. The mapping table stays inside the Cortex gateway. When the reply arrives, Cortex replaces the tokens with your original values: seeing your own data in a reply is the designed behavior, not a leak. The “× redacted items” chip appears on a reply only when the reply itself contained values matching a pattern; it does not mean that your values were restored.

Click one of your messages, or the “Protection panel” icon at the top right, to see what was protected, where the request went and which policy applied.

Detection relies on patterns, plus name recognition and detection by a local model when the AI engine is enabled: it covers the known formats of the displayed DLP profile, with no guarantee of exhaustiveness. Avoid pasting data you do not need.

History and preferences

“My history” (in the left column) lists your requests from the last 30 days: the time, the destination, the number of redacted items and the status of each. It also shows what Cortex keeps: your conversations are stored encrypted and visible to you alone, and the table of redacted values never leaves the gateway. The “Export my exchanges (CSV)” button downloads this log. Auditors see the same trace, without the content of your messages.

“Settings” applies to this browser only:

  • the interface language (Canadian French or English);
  • the default destination, among those your administrator allows;
  • the send key (Enter or Ctrl + Enter);
  • showing the model's reasoning, reducing animations and showing sources under replies.

The light or dark theme is changed with the icon at the top right of the chat. Security rules (destinations, DLP profiles, retention) are set by your administrator and are read-only.