A question in one sentence, the answer right away.

A simple figure costs a phone call, or an hour in a spreadsheet.

The answer to a management question isn't written anywhere: it gets reconstructed. Someone opens a spreadsheet, cross-checks two exports, calls whoever is handling the file. The figure comes out an hour later, and nobody can check it.

• A phone call to find out where an order stands
• A spreadsheet to open, filter and cross-check for a single figure
• An answer that depends on the one person who knows the file
• The document nobody can find the day a client disputes

It answers with your figures, not with generalities.

You type the question the way you'd ask a colleague. The assistant searches what your app holds — jobs, orders, clients, suppliers, documents — and returns the figure, the list or the document, along with the record it came from.

When a client disputes something, the dated photo and the signed document come out in one click, instead of being hunted down in an inbox.

• “Where does this order stand?”
• “How much did we pay this supplier last year?”
• “Which jobs are waiting on a signature?”
• “Who needs chasing this week?”

It doesn't guess: every answer points back to the record it came from.

An assistant that invents a figure is worse than a spreadsheet: you don't know it got it wrong. This one searches your app and nowhere else, and every answer carries the link to the job, the order or the document behind it.

It sees what the logged-in person is allowed to see, and no more. A site manager doesn't get the company margin because they phrased the question well.

• Every answer carries a link to its source
• It searches your app, not the web
• When the data isn't there, it says so instead of inventing it
• Role permissions apply to answers just as they do to screens