Doubling Grant Applications with GenAI — and a Disclosure Policy

The situation

A small environmental not-for-profit with two full-time staff and a heavy annual grant-writing workload was watching funders shift faster than the organisation could respond. In a single funding round, they wanted to submit more applications than in previous years, but the human capacity to draft, review, and submit was not there.

A staff member suggested using a free-tier generative AI tool for first drafts.

What they did

Before using the tool, the organisation set three rules: never paste funder-confidential information or beneficiary personal details into the AI; always conduct a human fact-check pass on every output; and disclose AI involvement to funders when asked.

With those guardrails in place, staff used the free-tier tool to draft initial grant applications, then spent their time on the higher-value work: fact-checking claims, inserting evidence specific to their work, and making sure the application sounded genuinely like the organisation.

In the relevant funding round, they submitted roughly twice as many applications as in the previous year.

What it cost

Zero. The tool used was a free-tier generative AI product. Staff time per application was the same or slightly less, with more applications submitted overall.

What the board did

The CEO brought the question of disclosure to the board. Not every funder asks about AI involvement, but some are starting to — and the board needed to decide what the organisation's position would be. The board resolved that the organisation would disclose material AI use when asked, and would note AI assistance in grant acknowledgements where funders requested transparency.

The board also asked management to log which funders specifically enquire about AI use, and set a twelve-month review date to revisit the disclosure policy as funder expectations evolve.

The board did not generate the guardrails — that was management's work. But setting the disclosure position was a values question that appropriately landed at board level.

The outcome

The organisation did not track a baseline before implementing AI assistance, so a precise funding uplift cannot be stated. Staff estimate they submitted roughly twice as many applications with the same headcount, including two grants they would previously not have had capacity to attempt. No funder has raised concerns about AI use. The disclosure policy has not needed to be tested in a difficult conversation.

Lessons for directors

• The question of whether to disclose AI use to funders is a values question, not just a communications question. It belongs at board level.
• Free-tier tools can deliver genuine value without budget exposure — but they need clear rules about what data goes into them.
• Doubling output without doubling cost is a meaningful result. AI can change the economics of grant-dependent NFPs in ways that deserve proper governance attention, not just management improvisation.
• Be honest about what you can and cannot measure. An honest estimate is more useful to a board than an invented figure.