Every number the assistant gives a CSM can now be checked
Until now, asking the assistant how much revenue was at risk got you a sentence with a number in it. The number was real — it came from your own live data — but nothing on screen told you what it counted, what it left out, or how it was reached. A figure you cannot check is a figure you have to take on trust, and a conversational answer reads more confidently than it has earned.
Numbers now arrive as claims. A claim is the figure plus everything you need to judge it.
What a claim shows you
Ask “how much revenue is at risk across my portfolio?” or “how many accounts am I carrying?” and you get:
- The figure, and the question it is answering.
- What it counts — the population it was taken over, whether it is a total for a period or a level right now, and which date decides whether something is in or out.
- What is not counted here, said plainly. A blocked account with no accepted quote yet has no value recorded against it, so it is left out and counted as left out — never quietly counted as zero.
- How it adds up. The subtotals are on screen and they reach the headline. If four accounts are red and two amber, you can see the red total and the amber total and that they sum to the number above them.
- The records behind it, one click away.
Where a figure was measured against the platform’s shared metric definitions, the claim shows those definitions in their own words. Where the figure is defined by the answer itself, it says so. You should never have to guess which you are looking at.
Money says what it cannot do
Quotes are raised in different currencies, and no exchange rate is recorded against them. Rather than converting at today’s rate to make a total look complete, the assistant leaves those amounts out of the total and tells you how many it left out and why. A total that quietly mixed currencies would be worse than a smaller one that is honest about its edges.
If a blocked account records an amount but no currency at all, there is no currency to total it in, and the assistant says there is no total to give rather than picking one.
Breaking things down, and where an account has got to
Ask to break your blockers down by stage, by CSM or by status and you get a table or a funnel built for that question — labelled as generated for it, so it is never mistaken for a saved report you could come back to next week. The parts always add back up to the whole, and percentages total 100.
Ask where an account has got to and you get its position on the go-live journey, with the stages named the way a CSM sees them, and any drop-off points shown separately rather than pretended to be stages everything passes through.
Seeing how an answer was reached
Under every answer that looked something up there is now “How this was answered”. Open it and you see which tools ran, which organisation each was bound to, what it was asked for and — where your access narrowed it — what it actually ran over, plus how long each took. If nothing was looked up, it says so, rather than showing you an empty list.
What has not changed
All of this is still read-only. The assistant can show you what is stuck and what it is worth; it cannot chase it, move a status or resolve a blocker. And if a figure ever arrives without the records that would reproduce it, the assistant shows it as plain text and tells you it could not be shown in full — it will not render a number you have no way to check.