'She seems confident' is unevidenced. This shows how to structure mentor session notes in Sopact Sense — extracting topics, progress, blockers, and actions with evidence, and never inferring emotion.
In short: Mentor notes go wrong when they record a feeling no one expressed. In Sopact Sense the extraction pulls topics, progress, blockers, action items, and any relationship signal — but only where a verbatim line supports it. 'She seems confident' with no quote behind it gets caught, not recorded.
Point the Assistant at clean mentor notes with persistent contact IDs so every session stays tied to the same mentee over time. Load your Decision Brief first — the decision, the audience, the fields you track, and your evidence standard — so the extraction fills the right structure and holds itself to evidence.
From the mentor note, extract topics, progress, blockers, action items, relationship signal with evidence. Don't infer unexpressed emotion. Grade green/amber/red.
The prompt carries five elements. The dataset is the loaded mentor notes. The fields are topics, progress, blockers, action items, and relationship signal. Evidence per field ties each value to a quoted line. No inferred emotion forbids reading confidence or mood into the note. Grade G/A/R rates how well-evidenced each field is.
Run on the Mentorship & LMS dataset (DEMO 09) already loaded in Sopact Sense.
GRADE: green | 1 | Explicit action; amber | 1 | Inferred emotion; red | 1 | Blocker not captured
Sense returns each field with its supporting quote. An explicit action item the mentor wrote down comes back green. A confidence read with no quote behind it comes back amber — it is an inference, not evidence. A blocker that was raised in session but never recorded comes back red: the field is simply missing.
Take the lowest-graded element above and fix it using only what the program could realistically measure. Show the before → after grade and the single indicator/edit that moves it to green.
For the missing blocker, the fix is structural: make blocker a required field in the mentor note template. That single change turns a silently dropped red into a captured, evidenced field every session.
Create a 'missing & incomplete' report from this analysis in Sopact branding [or paste your website URL / brand guideline to apply your own]. List every element graded amber or red, what is missing, and the one input that fixes each. Lead with the decision this report informs.
Create a shareable link for this report and open it in a new tab.
Quote or leave blank. A relationship signal without a verbatim line is a guess. If there's no quote, the field stays empty — that is the discipline.
Separate progress from feeling. 'Completed the draft' is progress; 'seemed proud of the draft' is inferred emotion. Keep them in different columns.
Make blockers mandatory. Notes that only capture wins quietly hide the things that need escalation. Require the field.
Watch for flattering drift. Models tend to round up toward positive readings.
Audit a sample of notes for any emotion word that has no quote behind it, and strip it before it reaches a report.
Extract a fixed set of fields — topics, progress, blockers, action items, and relationship signal — and tie each to a quoted line from the note. Sopact Sense does this in one prompt and refuses to infer emotion, so the record stays evidenced rather than impressionistic.
Because an unevidenced feeling like 'she seems confident' becomes data the program can't defend. Recording only what a quote supports keeps the note credible and protects the mentee from being mischaracterised.
Blockers that were raised in session but never written down. The fix is to make blocker a required field in the note template so it is captured every time.
Open Sopact Sense, paste your program description, and put it to work.
Try in Sopact
