Pooja, 30: a grant proposal, a funder's forty-page template, and a programme she knows better than anyone

Planned for release phase 2. The page is complete as a draft; interactive parts run in the browser and do not yet save to an account.

1. Step 1 — Define the End State

   "A proposal the funder's programme officer reads and thinks: these people know exactly what happens in these villages, and they can count. Every number is one I can defend in a field visit. The theory of change is what we actually believe, in their format. And the budget survives their finance team." Audiences: the programme officer, the funder's finance reviewer, her director, and a teacher in a named village.

2. Step 2 — Map the Flow

   Her list: understand the template, write what the programme is, map it into the sections, budget, review. AI review adds: read the funder's last three funded proposals in this area, which are public — the template is less the point than what they fund; a glossary of their terms mapped to her own; and a logframe built from her monitoring data, not from scratch. It suggests "have the tool write the first draft from the template". She declines; that is the consultant's proposal with a different author.

3. Step 3 — Assign Tasks

   What the programme does, believes, and has failed at: hers, written by hand before any template is opened. Explaining the funder's vocabulary: AI as tutor, with the template pasted in. Mapping her paragraphs to the template's sections: AI proposes, she decides. Turning her monitoring spreadsheets into the logframe's indicators: hybrid — the tool structures, she verifies every number against the field data. Budget arithmetic in the funder's format: AI, verifiable, checked by her accountant. The teacher's story on page one: hers, from a real visit.

4. Step 4 — Select the Tool

   A document tool with long context (template, three funded proposals, her own monitoring reports all loaded); a spreadsheet for the budget, never a chat tool; a conversational tool for the vocabulary lessons.

5. Step 5 — Prepare and Direct

   Criteria: every indicator has a data source I can name; every number reconciles with the monitoring sheet; no sentence I would be embarrassed to read aloud to a village teacher; the theory of change contains one thing we tried that failed; the budget totals match to the rupee in both formats.

6. Steps 6 and 7 — Execute and Evaluate

   The vocabulary lessons take a day and change everything — she had been reading "outcome" and "output" as synonyms. The tool's section mapping puts her failure paragraph in an appendix; she moves it to the theory of change, because the criterion says it belongs there and because the three funded proposals all had one. The logframe draft invents a baseline where she has none; the criterion catches it and she writes "baseline to be established in month one", which is true.

7. Step 8 — Integrate and Stress-Test

   Assembled, read as the programme officer in thirty minutes. Small fixes: two acronyms undefined. Structural: the budget and the logframe disagree on the number of villages — a Step 3 mismatch, two tasks built from two versions of the plan. Reconciled to the current one. Her director reads it and, for the first time, recognises the programme in a proposal.

What changed

Funded, with a note from the programme officer about the honesty of the failure paragraph. The template did not swallow the programme this time, because the programme was written down first, and the template was learned as a language rather than obeyed as a form.

A composite case, not a client engagement; the person is invented, the situation is not.