Sopact is a technology based social enterprise committed to helping organizations measure impact by directly involving their stakeholders.
Copyright 2015-2026 © sopact. All rights reserved.
Score any grant proposal against a weighted rubric in seconds — and catch the high marks that rest on no evidence. Here is the exact prompt and what Sopact Sense returns.
In short: To score a grant proposal with a rubric, load the application into an AI workspace, give it your weighted criteria (need-evidence, capacity, feasibility, impact, budget, sustainability), and require it to validate every claim against the attachments and cite the field behind each score. The point isn't the number. It's whether each score has a field behind it. Sopact Sense grades each element green, amber, or red so you fund on proof, not on prose.
Start with the grant round already loaded as clean data with persistent contact IDs, so every claim ties back to a real application field and attachment. Point the assistant at the dataset and have it read your Decision Brief first — the decision, audience, outcomes, indicators, and evidence standard — so it scores against your bar, not a generic one.
You are the Sopact Sense Assistant working over the DEMO-05 · Grant Applications dataset (clean data + persistent contact IDs). Load my Decision Brief (decision, audience, outcomes, indicators, evidence standard) first, then wait for my task.
The scoring prompt names the rubric and forces evidence. Paste this verbatim:
Score proposal [ID] on need-evidence, capacity, feasibility, impact, budget, sustainability (1-5, weighted). Validate every claim against the attachments; cite the field behind each. Grade green/amber/red.
The prompt works because of five elements: the dataset it scores over, the rubric and its weights, the instruction to validate claims against attachments, the rule to not reward unsupported claims, and the demand to grade green/amber/red so weak links are visible.
Run it against the Foundation grant round demo:
Run on the Grant Applications dataset (DEMO-05) already loaded in Sopact Sense.
GRADE: green | 62% | verified outcome; amber | claim | unevidenced impact; red | budget | missing detail
The green element is a verified 62% outcome figure with a field behind it. The amber element is an impact claim that scored well but has no attachment to back it. The red element is missing budget detail — there is no line-item table to validate feasibility against.
Take the lowest-graded element and fix it with the one input that moves it. Sense shows the before → after grade and the single edit that does the work.
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.
Turn the analysis into a reviewer-ready report, then a link that opens with no login.
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.
Score the evidence, not the writing. A polished narrative often hides a missing number. Always require a cited field behind each score so persuasive prose can't earn green on its own.
Weight before you score. Set the rubric weights up front. If feasibility matters more than sustainability for this round, say so in the prompt — otherwise every element reads as equally important.
Flag, don't fail. A missing budget table is a request, not a rejection. Mark it red and ask for the one input that fixes it rather than scoring the whole proposal down.
Re-score after the fix. Once a weak element is repaired, re-run only that element so your final score reflects the corrected evidence.
Re-score only the budget element for proposal [ID] now that the line-item table is attached, and show the updated weighted total.
List the dimensions that drive your funding decision — typically need-evidence, capacity, feasibility, impact, budget, and sustainability — assign each a weight that reflects its importance for this round, and score each 1–5. The critical addition is an evidence rule: every score must cite the application field or attachment behind it, so a claim with no support can't earn a high mark.
AI scores reliably when it is constrained to your rubric and forced to cite evidence. The risk is rewarding fluent writing; you remove it by requiring a cited field per score and grading green/amber/red so unsupported claims surface as amber instead of disappearing into the total.
A score is weak when high marks rest on unevidenced claims or when a key element — like a line-item budget — is missing entirely. The fix is to flag those gaps, request the one input that resolves each, and re-score, rather than letting prose carry the proposal.
Open Sopact Sense, paste your program description, and put it to work.
Try in Sopact