AI for Grants, Proposals, and Donor Readiness
This workbook helps proposal teams use AI as a support tool for structure, clarity, review, and planning. AI must not become the source of evidence, the final compliance authority, or the owner of donor-facing claims.
1. Donor Call Analysis
| Item to extract | What to record | Human check |
|---|---|---|
| Eligibility | Applicant type, country, partners, registration, consortium rules | Check against original call text |
| Priorities | Sectors, target groups, cross-cutting themes, geographic focus | Confirm exact donor wording |
| Compliance | Page limits, forms, annexes, budget rules, deadline, submission portal | Assign one compliance owner |
| Scoring logic | Evaluation criteria, weighting, mandatory requirements | Build a review checklist |
2. Evidence Discipline
Before using AI to draft a problem statement, separate evidence from assumption.
- Known evidence: verified data, needs assessment findings, official reports, field notes with date and source.
- Reasoned assumption: plausible but not yet proven interpretation that must be labelled and checked.
- Do not use: invented numbers, unattributed quotes, unsupported beneficiary claims, or copied donor text.
3. Safe Prompt Pattern
Use the notes below only. Do not add statistics, donor requirements, partner names, or claims that are not in the notes. Produce a draft paragraph with three labelled parts: evidence, interpretation, and proposal relevance. End with questions a human reviewer should answer before submission.
4. Logframe Review
| Question | Review note |
|---|---|
| Does each activity connect to an output? | |
| Does each output support the outcome? | |
| Are indicators measurable with available capacity? | |
| Are targets realistic for time, budget, access, and staffing? | |
| Are assumptions explicit? |
5. Final Quality Gate
- Every factual claim links to a real source.
- The proposal language reflects the organization, not generic AI style.
- The budget narrative matches activities, staffing, procurement, and timeline.
- Safeguarding, accountability, data protection, and localization are not decorative add-ons.
- A named human lead signs off on the final donor-facing version.
6. Source Basis
This starter material is informed by public responsible-AI and humanitarian data responsibility guidance, including OECD AI Principles, NIST AI Risk Management Framework, IASC Operational Guidance on Data Responsibility in Humanitarian Action, and public donor practice around evidence, compliance, and accountability. It adapts those ideas into a practical proposal-training format.