Check a contract or grant agreement
Read a funder's agreement or a supplier contract against your own policies and red lines, and flag the clauses that need a closer look, each one cited so your trustees can see exactly where it came from.
For UK charities · More private AI use cases
What else can private AI build for your charity?
The use cases on this site are starting points, not the limit.
Each detailed page here answers one common question. But the idea underneath them all, your data analysed or your documents worked with inside your own systems, applies to almost anything you'd otherwise be tempted to paste into a public chatbot.
Here are more of the things charities ask me to build, and a simple test for whether your own idea fits.
How to tell if your idea fits
Almost everything on this site comes down to one of two patterns, both running on infrastructure you control rather than a stranger's.
The first is analysing your own data: an assistant sits next to a read-only copy of your records and answers plain-English questions with figures, tables and charts, showing the query behind every answer. The second is working with your own documents: an assistant reads your policies, contracts, applications or notes and answers, summarises or drafts, with every claim cited back to the source so you can check it.
So the test is simple. If there's a task where you'd reach for ChatGPT but the data behind it is too sensitive to paste in, it can almost certainly be built privately instead. Not sure whether you even need AI for it? That's a fair question too, and worth asking first.
Work with your documents
The same build behind writing funding bids, summarising case notes and searching your documents, pointed at other everyday jobs.
Qualify a tender or funding opportunity
Check an opportunity against your eligibility, your priorities and your capacity before you spend days on a bid, so you only chase the ones that are genuinely worth it.
Answer "what does the regulator actually say?"
Ask official guidance, from the Charity Commission, the Fundraising Regulator or the ICO, a plain-English question and get the answer with the source, instead of hunting through PDFs.
An everyday staff assistant
Answer the recurring "how do we do this here?" questions from your own handbook and policies, so new staff and volunteers can get unstuck without interrupting someone else.
Turn notes into minutes
Turn rough notes or a recording of a meeting into clean minutes and a clear list of actions, in your own format, ready to check and send.
Analyse your data
The same build behind analysing donor data and membership data, pointed at the other numbers your charity has to make sense of.
Outcomes and impact reporting
Pull the figures funders ask for from your own service data, so the same quarterly and annual reporting questions get answered in seconds rather than rebuilt by hand every time.
Finance and budgets
Ask plain questions of your finance data: spend against budget, how restricted funds are tracking, where the money actually went, without exporting it to a spreadsheet and a public tool.
Volunteer and survey data
Make sense of volunteer records, sign-ups or survey responses, who's active, what's changing, what people are telling you, on data that never leaves your control.
These are still only examples. The list of jobs this idea fits is far longer than any page can hold.
Got something else in mind?
If there's a task your team keeps reaching for ChatGPT to do, or one you've held back from because of where the data would go, I'd like to hear about it. Tell me the job and the data behind it, and I'll tell you honestly whether it's a good fit for private AI, whether there's a simpler way, or whether you don't really need AI for it at all.