Skip to main content
You do not need to read this to use Exec through an assistant. Ask in plain language and it works out which capability to use. This page exists for the reader who wants to know exactly where the edges are, which is usually somebody doing a review. Everything below is subject to your own Exec permissions. Having a capability listed here does not mean you can use it on data you cannot otherwise see. See What the AI can see.

​
Reading your workspace

​
Practice and results

Ask for feedback across many sessions in one go rather than one at a time. It comes back in a single request, so ten sessions is about as fast as one. Save the full transcript for the one session you want to examine closely.

​
Skills and scenarios

These require workspace admin or group admin, including when a member asks only about their own scores. Their own sessions, grades, and coach feedback are still readable through the rows above.

​
Assignments

Status values are not started, in progress, completed, past due, and did not pass. Each row carries the due date, the best score and rank so far, and attempts against what was required. Assigning needs the same permission it needs in the product, and the assignment carries your name, the same as one you create in Exec.

​
Building a roleplay

Building takes anywhere from under a minute to a few, and your assistant checks back on its own. It always describes the draft before publishing. Stopping part way keeps whatever was already written to the draft. Publishing enforces your workspace’s limit on custom scenarios, so if you are at your limit it will say so.
Publishing shares the scenario with the whole workspace. To keep it to yourself, ask to set its visibility to you.
One quirk worth knowing: scenario requests are checked for content that reads as code or as instructions to the assistant, and those are rejected with a note asking you to rephrase. If you paste a formatted scenario spec and it is refused, describe it in plain sentences instead. That is the most common surprise in this whole feature.

​
Courses

Enrolling works on published courses and follows the same rules as the course’s Participants tab in Exec, so the person enrolled gets the same email they would get from the product.

​
People, groups, and seats

Each of these needs the same permission the action needs in the product, so a member without admin cannot invite people or change seats through an assistant either.

​
Charts and dashboards

Charts come out in Exec’s colours and use Exec’s vocabulary. They need an assistant that can display interactive panels. In a text-only assistant you get the numbers instead, and nothing breaks. See Limits.

​
Recorded calls

What comes back is whatever your scorecards and call categories are configured to capture, not a fixed set of measures. A question about objection handling lands if objection handling is one of the things you score. If it comes back thin, ask in the language of your own criteria. See Call scoring for how categories and scorecards are set up.

​
Knowledge Hub

You see the hubs you can open in Exec. The same material can be used as source material when building a scenario.

​
Everything else in Exec

Certifications and programs have no dedicated capability, but an assistant can still query them, so questions like who is certified on the discovery call scenario? generally work. Expect an extra step or two while it works out how to ask.

​
How changes work

Every change on this page is built to be previewed first. Your assistant runs it as a dry run that changes nothing, shows you the result, and runs it for real once you confirm, so the habit to keep is reading the preview before you say yes. A change you lack permission for in the product is refused in the same way. Changes always run as you and carry your name, which is why a workspace API key can read but cannot change anything: sign in with your own account to assign, enroll, or invite.

​
Next

What the AI can see

Permissions, in full

Limits

Sizes, caps, and what to do about them