“Only the eastern region, please.”
The chart covers every region because that is what the board is for. You are responsible for one of them. The agent answers the same question for your slice, in your session.
AI Agent
A dashboard answers the question it was built for. The one you ask next is usually your own – one region, one month, one customer. Put it to the agent in the same view and the answer arrives on your screen, while the board everyone else opened stays exactly where it was.
30 days free · No credit card required
What a follow-up looks like
None of these needs a new dashboard, a request to a colleague or a copy of the board with one filter changed.
The chart covers every region because that is what the board is for. You are responsible for one of them. The agent answers the same question for your slice, in your session.
Comparisons are the most frequent second question. The period shifts in the answer you get back – not in the filter your colleagues are working with.
A dip in a line is a starting point, not a finding. Ask what sits underneath it and the agent takes the drill-down step for you, as far down as the model goes.
Some answers are a list, not a picture. Ask for the rows behind an aggregate and you get them as a table you can read through and export.
The forecast from your own history, seasonality and trend included, can be set beside the budget figure in the answer – without anything changing on the board it came from.
Reword it, swap one field, switch from English to German. Each attempt costs a few seconds and leaves no trace on the dashboard.
The boundary
The agent works from the dashboard in front of you: the fields on it, the model behind them and the filters that happen to be set. It reads that model rather than a copy of your rows, and the figures it needs are calculated where the data already sits.
It also sees no more than you do. The permissions on your account apply to its answer exactly as they apply to the board – ask about margin as a regional manager and you get the margin of your region, because that is what the query returns for you.
A follow-up answered in words is the case where figures travel: the aggregated values the answer quotes go out with the question, the way they stand on the board. The records behind them do not. A team that would rather work without the agent can have it switched off for the environment, and then nothing goes out at all.
Your follow-up lives on your screen. The saved dashboard keeps its widgets, its layout and its filters, so the colleague who opened the same address two minutes ago still sees what they were shown.
The data underneath is equally untouched. A question stays a question: nothing is written back to a source, no record is corrected, no scheduled refresh moves. Leave the view and the detour goes with it.
An answer worth keeping does not have to be lost, either. Put it on a board of your own or pass it on as a link – then it becomes something people can return to, and the agent goes back to answering the next question.
In a demo we open a dashboard with your data and put your follow-up to the agent, live.