There is no shared, platform-wide key any more. If your organisation has not configured its own,
there are no AI features anywhere in the product — not here, not on a work order, not on a create
form.
Where you use it
There is a Copilot tab on every work order and every asset, and that is where it is most useful: the place to ask what is wrong with a pump is on the pump. A “fill from a document” button and a voice option, described below, also draw on it directly from create forms. The Copilot activity screen under Analytics is history and effectiveness only — what has been asked, what got accepted, and the accept-rate-per-capability material in Is it any good. It has no switches on it. Whether Copilot is on, whose key it spends, which capabilities are allowed and how many questions a day are all set on Administration → AI instead.If Copilot is off, Copilot activity says so and says why: it runs on this organisation’s own API
key, and points an administrator at Administration → AI rather than offering a
switch itself.
What you can ask for
Which of these your organisation has agreed to is set on Administration → AI. Anything
not agreed to is refused by the server, not merely hidden from the menu.
Asking
Pick a capability, type the question, and ask. A counter shows how many questions your organisation has left today — the same ceiling that is configured on Administration → AI. The answer comes back through a queue, so there is a gap between asking and being answered — the screen says so rather than showing a spinner that implies something is happening faster than it is. A conversation stays on the record it was opened against, so navigating away and coming back does not lose it.Turns that failed or were declined
They appear in the conversation, in place, with the reason. A conversation that quietly dropped the answers nobody liked would read as a product that works better than it does.What was sent
Every question you ask carries a What was sent link that opens the record exactly as it stood when it left. It is one click away on every turn, not buried in a policy document.Saying whether it was any good
Useful, Wrong and Unsafe are on every conversation. The last two are the ones worth reading.Suggestions
The copilot proposes; it does not write. A suggestion sits there until somebody decides on it, and nothing reaches a record on its own. Each suggestion shows the field, the proposed value, the reasoning, what it rests on, and the confidence the model gave itself — presented as the model’s own number rather than dressed as a measurement. Three things you can do with one:
The middle option exists for a real situation: a technician can agree with a diagnosis without being
the person who classifies failures.
Asking and accepting are separate permissions, because sending a record out and writing to one are
different risks. By default technicians may ask and read; supervisors, planners, managers and
administrators may also accept. See roles.
The Waiting tab on Copilot activity collects every suggestion across the organisation that
nobody has decided on yet.
Filling a form instead of typing one
Two ways to start a record from something other than a keyboard, available on Create forms — an asset, for instance — wherever your organisation has Copilot enabled with the right capability. Neither is available on edit: they exist to get a new record started, not to rewrite one that already has a history.Fill from a document
A Fill from a document button on the form: photograph a nameplate, or upload a PDF or image, and matching fields fill in automatically, right there in the still-unsaved form. Each field it touches carries a small “read” (sparkle) marker until you edit it, so you can tell what the machine told you from what you typed yourself. Nothing is saved until you press Create — reading a document only ever proposes values. Once you do save, the source photo or PDF can attach itself to the new record automatically, so the evidence travels with the row it produced. This needs Copilot enabled with the Extract capability. Without it, the button is not there.Voice — “describe it instead”
A microphone button beside it does the same job by ear. Record yourself describing the item — “it’s a KSB pump, model…” — and the recording is transcribed and run through the same fill pipeline. The transcript is shown above the fields, so you can see what was actually heard rather than trust that it matched what you said. Voice fill needs its own OpenAI API key, configured separately under Administration → AI, because Anthropic has no speech-to-text capability of its own. An organisation running Copilot entirely on an Anthropic key gets everything above except the microphone — add an OpenAI key for that one piece and the button appears. If no such key is set, the button simply is not there.The recording itself is never stored. What stays is the transcript text, kept on the conversation
like any other question.
Two smaller behaviours worth knowing
- A picker field filled this way shows the filled value with a Loading… label rather than going blank while its option list resolves in the background — the value is already decided, the picker is just catching up.
- A picker that is scoped to another field’s value — a state that depends on a country, say — is never offered to fill automatically. Filling one half of a dependent pair without the other risks a mismatch neither the model nor you would notice until later. Those fields stay manual, on purpose.