Back to profile

Case Study: Nonprofit Proposal Go/No-Go Engine

TL;DR

Evidence

Metrics

Context/Constraint

Nonprofit teams rarely fail because they cannot write enough text.

They fail because the proposal logic is thin, donor fit is assumed, evidence gaps are hidden, or the budget/timeline does not survive review.

Under deadline pressure, teams need a hard read before drafting turns into sunk cost.

Problem

Most proposal-assistant workflows optimize for narrative production. That is useful, but dangerous when the underlying submission is weak.

A polished proposal can still be a bad submission if it has unsupported targets, unclear assumptions, weak donor alignment, safeguarding gaps, or a budget that does not match the intervention.

Actions

Fact → Assumption → Hypothesis → Unknown → Verdict

What it does now

Current outputs

Who it is for

Why this version is better

It does not try to make weak proposals sound strong. It exposes the weaknesses early.

That is the point. A Go/No-Go engine is more useful than a drafting assistant when time, credibility, and donor fit matter.

Tech stack

Relevance

This project shows practical agent design for nonprofit operations: turn ambiguous inputs into a structured decision, preserve uncertainty, and prevent polished text from hiding submission risk.

Project links

Author: Vassiliy Lakhonin