What the work involves
You write reference artifacts and judge model output against them. A task might be: draft a two-page LOI narrative for a $250K capacity-building request, build a theory-of-change logic model in a spreadsheet with outputs and outcomes properly separated, assemble a board finance-and-program briefing deck, or take a model-generated donor impact report and mark exactly where it overstates attribution, invents a beneficiary count, or blurs restricted and unrestricted funds. Much of the work is comparative — two AI drafts of a community needs assessment, and you explain in writing which one a program officer would actually accept and why. Rationales matter as much as ratings; the lab is training on your reasoning, not just your score.
What the platform screens for
- Real funder-facing authorship, not adjacent exposure. Screeners probe for specific documents you personally drafted, who reviewed them, and what the funder said back.
- Sector fluency under follow-up: logic model vs. theory of change, outputs vs. outcomes, Form 990 basics, fiscal sponsorship, restricted vs. unrestricted revenue, indirect cost recovery, DAF versus private foundation mechanics.
- Craftsmanship — clean tables, formulas that hold up, decks built to be read in a board packet rather than narrated.
- Evaluation judgment: can you separate a document that is merely fluent from one that is factually and structurally sound, and articulate the difference in a paragraph someone else could act on?
Logistics
Fully remote and asynchronous. Work is drawn from a queue on your own schedule; 5–20 hours a week is the stated band, with more available when volume allows. There is an AI voice screen — a live, structured conversation where follow-ups adapt to your answers, so rehearsed summaries hold up poorly. The $70/hour figure is what has been observed on this listing, not a guaranteed rate; pay and volume can shift with project phase.