What Ace Can Reach
Ace does not have general access to your database. It reads through a defined set of capabilities — about seventy — each of which does one specific thing, such as fetching point-of-sale events or looking up a subject's detections.
Knowing roughly what those cover tells you what is reasonable to ask.
By module
| Area | Roughly what it can read |
|---|---|
| Faces and plates | Subjects and their detections, demographics, tags and tag statistics, zones and zone visits, alerts, detectors, licence plates and their detections, similar-plate matches, subject-to-plate associations, saved queries, aggregation, report generation, and face search from an attached image |
| Casino | Events, ratings and rating groups, rating summaries, the gaming floor, shift reports, bet sessions, TITO tickets, roulette, alerts, message events, timelines, player appraisals, entities, game types, saved queries and aggregation |
| Point of sale | Events, entity statistics, entities, saved queries and aggregation |
| Baccarat | Game results and the baccarat data behind them, where the module is licensed |
| Cameras | The device list, individual devices, device groups, camera associations, and opening cameras for you |
| Usage statistics | Sessions, session detail, sign-in summaries, plugin events and saved queries |
| Core | Installed modules, server information, navigation and opening views |
It can also produce a PDF, send an email at your request, refine a result it has already fetched, set a chat's title, and remember a preference you tell it.
Baccarat is worth calling out because it was the gap people noticed: asking about baccarat results used to be answered from the general casino data, which is a different set of records. Ace now reads baccarat's own, so the figures it quotes are the ones the Baccarat module shows.
What this means in practice
Ask about things it can read. Detections, transactions, ratings, sessions, tags, zones — all well covered. Questions about data eConnect does not hold are outside what any amount of phrasing will fix.
Aggregation is available across the main modules, so "how many", "which is highest", "break that down by" are natural questions rather than special ones.
It can use your saved queries. If you have already defined the thing you care about, the assistant can work from it instead of re-deriving it — one good reason to keep saved queries tidy and well named.
Gated by licence and permission
Two filters sit in front of everything above.
A capability belonging to a module you are not licensed for is not available at all. A capability you lack permission for is not available to you, even though a colleague may have it.
This is why two people can ask the same question and get different answers, and why an assistant that seems less capable than a colleague describes is usually a permissions difference rather than a fault. The full picture is in What Ace Can and Cannot See.
Reading vs. doing
Everything in the list above reads. The assistant does not edit subjects, change tags, alter transactions or modify configuration.
Two capabilities produce something outside eConnect — creating a PDF, and sending an email. Both act on your behalf and the email is confirmed with you before it is sent. See Reports and Email.
For administrators
Which capabilities are available, and how often each is used and fails, is visible under AI Usage and Analytics. A capability with a high error rate is the fastest explanation for answers that seem thin in one particular area.