Ask Cequence

Ask Cequence answers questions about your own AI Gateway tenant in plain language — what changed, what agents are doing, what data governance caught, how to get something done. Answers are built from your live data and rendered as charts and tables rather than walls of text, and anything that would change configuration asks you first.

Quick Start: What do you want to do?

I want to...	Go to
Open it	Open Ask Cequence
Know what to ask	What you can ask
Understand what it can change	What it can change
Find out why I cannot see it	Who can use it

---

Open Ask Cequence

Ask Cequence is not in the left sidebar. It is the button in the top bar, on every page, and it has a keyboard shortcut:

• Select Ask Cequence in the top right, or
• Press ⌘K (Mac) or Ctrl+K (Windows and Linux)

On a tablet or phone it appears as an icon in the app bar.

If the button is not there at all, you do not have access — see Who can use it.

Who can use it

Ask Cequence is limited to roles that can already see protection data across the tenant: Super Admin, Security Admin and Auditor.

Platform Operators, Network Admins and Employees do not see the button, and the keyboard shortcut does nothing for them. This is the most common reason someone cannot find the feature after a colleague mentions it.

Within that, Ask Cequence sees exactly what you see. It reads your tenant under your own identity, so it cannot show you anything your role would not already allow.

What you can ask

The panel suggests starting points grouped by who tends to ask them.

Group	Examples
Security & governance	What changed in the last 24 hours? · What sensitive data did DLP catch this week? · What AI traffic is running outside the gateway? · Which personas request tools they never call?
Operations	Which deployments are unhealthy right now? · Which MCP servers carry the most traffic, and is anything failing? · Which agents and personas were most active this week?
Build & connect	Which MCP servers and tools can I use to… · How does authentication work for an Agent Persona? · How do I connect Claude Code to an Agent Persona?
Help & how-to	How do I set up Data Governance for the first time? · Which IP addresses should I allowlist for the gateway?

Two kinds of question work well:

• About your tenant — answered from your live configuration and activity, including agent activity, deployment health, MCP servers, personas, and data governance policies and findings.
• About the product — answered from this documentation site, with the pages cited so you can read further.

What it can change

Almost everything Ask Cequence does is read-only. The exceptions are Data Governance policies, which it can create, update and delete.

Every one of those stops and asks first. When it proposes a change you see a Review this change card showing:

• exactly what will be changed, with the specific values
• how many personas are affected, and how many tool calls fall in scope
• similar findings in the last 24 hours, and any existing rules on the same scope

You can Approve and apply, Edit the proposal first, or cancel. Deleting requires typing delete to confirm — approving stays disabled until you do.

Nothing is applied without that confirmation, and every applied change lands in the audit log like any other configuration change.

Building an agent persona with it

The same assistant also powers the Agent Persona builder, at Agent Personas → New. Describe what the agent needs to do in plain language and it drafts a persona with the apps and tools to match, which you review before saving. See Agent personas.

That conversation is separate from the main Ask Cequence thread and does not carry over.

Tips

• ⌘K from anywhere. You do not need to navigate first — it works on every page.
• Name a time range. "This week" or "the last 24 hours" produces a much more useful answer than leaving it open.
• Ask it how to do things, not just what happened. It answers from the documentation and cites the page, which is often faster than searching.
• Read the blast radius on a change card. The number of affected personas is the quickest sanity check that a proposed policy is scoped the way you meant.
• It cannot see more than you can. If an answer looks incomplete, check whether your role has access to that area.