Schedule
Proof Points runs only when you ask. It has no schedule: it does not watch your sources, extract evidence overnight, or refresh the library while you are away. Each step starts with a person.
What Happens When You Ask
| You do this | Proof Points does this |
|---|---|
| Run the extraction command in Slack with a source | Reads the source and saves new candidates, each with a type, category, attribution, and confidence. When your own source names a client, it also writes a public version |
| Run the scrutiny command in Slack, or click Run Scrutiny in the app | Dedupes near-duplicates, asks a fast model to keep or cull the survivors, then verifies, scores, and ranks the ones it keeps |
| Approve a candidate in Needs Review | Makes the point available to your drafting agents |
| Reject a candidate | Archives the point and saves your reason; later extractions for the brand avoid that claim and similar ones |
Extraction and scrutiny take a few moments. Extraction starts in Slack, because the app screen has no place to enter a source. Scrutiny can be started from either.
Why It Works This Way
A proof-point library is only as trustworthy as the judgment behind it. A human chooses what evidence to capture and approves what becomes usable, so nothing is added on a timer and nothing is cited that you did not release.
Keeping the Library Current
Because nothing refreshes automatically, the library grows when you feed it. After a client call produces a strong result, or a new case study lands, run the extraction command with the summary, then review the candidates in the app.