A finding cannot cite what it did not read.
A single source can raise a finding. adduce grades how sure it is, and cites every record behind it.
A number moved. Three explanations sound right. Nobody has checked them against the records.
The dashboard shows the drop. The reasons live in the conversations, the tickets and the errors around it.
What goes in.
Only what Connect has already read into the project, plus the question you put to it.
Put one focused question to the project.
adduce tests each explanation against your records.
It reads what Connect brought in, nothing else.
What comes out.
An investigation tests the explanations against your records and says which held.
A finding you can question.
Each explanation is tested against your own records, not argued from the dashboard.
A grade for how sure it is.
Supported, refuted or inconclusive, with the evidence behind the verdict.
Every claim cites its record.
Open any claim and the row, the quote or the event it came from is there.
Limits and controls.
adduce enforces each line by rule, not by a prompt.
Citations are checked.
adduce never stores a finding’s citation that does not resolve to one of your own records.
Quiet weeks say so.
A quiet week is reported as quiet. The weekly narrative is still sent.
Severity stands apart.
How much it matters and how sure adduce is are separate grades.
Put your next finding on the record.
Connect one source. Ask the question your team needs answered, and read the grade adduce gives each finding.