Match quality is readable from the notes before a mentoring relationship breaks down. Here is how to assess fit and risk for a pair — and decide continue, re-match, or human review — with Sopact Sense.
In short: Match quality is readable from the notes long before a mentoring relationship breaks down. Point the Sopact Sense Assistant at every note for a pair, assess consistency, alignment, rapport, and any escalation flags, and get a clear recommendation — continue, re-match, or send to human review — graded so you can act before the quarter ends.
Tell the Assistant which dataset it is working over and load your Decision Brief first, so the fit read ties back to the decision, audience, outcomes, indicators, and evidence standard you already set.
You are the Sopact Sense Assistant working over the DEMO-09 · Mentorship & LMS dataset (clean data + persistent contact IDs). Load my Decision Brief (decision, audience, outcomes, indicators, evidence standard) first, then wait for my task.
For pair [PAIR_ID], assess fit and risk from all notes: consistency, alignment, rapport, escalation flags; recommend continue/re-match/human review. Grade green/amber/red.
The prompt carries five elements. Dataset: all notes for the pair. Fit factors: consistency, alignment, and rapport. Escalation flags: any safeguarding-style cue in the notes. Recommend plus human-check: continue, re-match, or route to human review. Grade: a green / amber / red call on the match.
Run on the Mentorship & LMS dataset (DEMO-09) already loaded in Sopact Sense.
GRADE: green | Rapport | ok · amber | Low | consistency · red | Escalation | flag
The Assistant reads every note for the pair and grades the fit. Green is healthy rapport — the relationship is working. Amber is low consistency: sessions are sporadic with long gaps. Red is an escalation flag that must go to human review before any "continue" call is made.
Take the lowest-graded element and fix it with something the program could realistically track.
Take the escalation-flag element above and fix it: route the pair to human review and log the outcome. Show the before → after grade and the single edit that moves it to green.
Create a 'missing & incomplete' report from this analysis in Sopact branding. 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 & open it in a new tab.
Never let the model auto-continue past a red. An escalation flag should always force human review, no matter how strong the rapport reads. Keep "human review" in the recommendation so a person signs off before a flagged pair continues.
Separate consistency from rapport. A warm pair that rarely meets is a different problem than a cold pair that meets weekly. Grading consistency on its own tells you which lever to pull.
Read all the notes, not the latest one. Fit is a pattern over time. Pointing the Assistant at the full note history catches a slow drift that the most recent note alone would miss.
Recommend the action, not just the score. Continue, re-match, or review is what a coordinator can use.
Ask for a one-line recommendation per pair so a coordinator can triage a whole caseload in minutes.
Read every note for a pair and assess consistency, alignment, rapport, and any escalation flags, then turn that into a clear recommendation — continue, re-match, or human review. The Sopact Sense Assistant does this in one pass and grades the match green, amber, or red, so you can spot a failing relationship from the notes before it breaks down.
Any escalation flag — a safeguarding-style cue or risk note — should grade red and route the pair to a person, regardless of how good the rapport looks. The recommendation should never auto-continue past a red.
Not on its own. Low consistency grades amber and signals a scheduling or fit problem to investigate; the fix is usually to track scheduled-versus-held sessions so the pattern is visible before deciding to re-match.
Open Sopact Sense, paste your program description, and put it to work.
Try in Sopact
