01
Governable
DPIA, trustee brief and consent framework in place from day one. Built for charity governance, not around it.
More mission,
less admin.
The agentic management system for UK charities. Safe AI, built with governance, audit and compliance from day one.
See the software work
One task, end to end.
Messy inputs in. A structured record out. A person signs off. Every figure traces back to source.
Step 1
Messy reality comes in
Emails, voicenotes, PDFs, surveys: whatever your team actually gets.
✉Email
“We ran 14 sessions this quarter, 92 unique participants.”
Voicenote · 1:48
“…the Tuesday group’s grown, must be forty regulars now…”
▤Q3-finance.pdf
Restricted spend reconciled at £74,210.
▦Beneficiary survey
n = 63, 88% reported improved confidence.
The problem
The work isn’t writing. It’s coordination.
30%of charities say their data systems are poor or non-existent, so moving and reconciling the numbers falls to people, by handSource: Charity Digital Skills Report, 2025
Pulling figures out of separate systems, reconciling spreadsheets, chasing the version that’s current. The writing is the quick part. The coordination is the week.
It’s a pattern right across charity work. The time that could go on donors or beneficiaries goes instead on moving data between systems and gluing them together by hand. That isn’t only slow, it’s unreliable: work has to be checked, verified and often redone. Data sits in the gaps between tools, and accountability rests on whoever happens to be holding it in their head. The result is uncertainty, and decisions made late or made tentatively.
What we build in
“Where did this number come from?”
The question every trustee, funder and auditor asks. The software has the receipt every time - and every figure that comes out is one you can stand behind.
Restricted spend this quarter: £74,210
- Traced to sourceRead from your finance export, Q3-finance.pdf, line by line.
- Reconciled deterministicallyThe quarter’s spend ties out to the penny. That check is code, not a judgement call.
- Ground-truth testedThe way figures are pulled is tested against numbers we already know are right.
- Signed offPriya, the Grants Manager, confirmed it before it went anywhere.
Checkable by you, line by line. We don’t claim zero errors - no honest tool can. We make every figure defensible.
Each task runs at the level of autonomy it needs. The assembly can run unattended, every quarter and every funder; anything that goes to a funder or a donor waits for a person.
Who we work with
The pattern
Faster. No corners cut.
The work that absorbs your team in coordination - buried in notes, inboxes, forms and spreadsheets - done at the pace the organisation needs.
Funder reports in a morning. Casework triage that catches what's missed. Supporter communications drafted from what your supporters actually said. Different tasks, the same software underneath.
Your data
Your data, yours, and resident in the UK.
A dedicated, isolated deployment we run for you. Your data lives in the UK, is used only for your work, and stays yours - the data and the code, exported in full whenever you like.
- A dedicated, isolated deployment we run. Its own project, separate from every other charity's - not a shared, multi-tenant tool.
- UK data residency. Persistent data stays in the UK. We connect to your source systems to read what a task needs; we don't co-mingle your data with anyone else's.
- Never used to train a model. Yours stays yours.
- Portable, not proprietary. Outputs in open formats; the configuration and the code are yours to take with you. No lock-in.
Safe by design
Built for charity governance, not around it.
02
Auditable
Every output traces back to its source. The audit trail is part of the product, not an afterthought.
03
Compliant
Your data stays in the UK, used only for your work. Never used to train a model.
04
Human-led
A person signs off at every step that matters. Nothing leaves without your decision.
Who’s behind it
Three years building production AI for UK charities.
Make Sense Of It has spent three years building production AI for UK charities. Different tasks, the same shape underneath: messy inputs in, a defensible result out, a person signing off.
Breast Cancer Now
Every patient’s voice, as evidence
Frontline feedback from breast cancer services, turned into evidence the team can act on - every voice, not a sample of them.
Read the case study →National Lottery Heritage Fund
A thinking partner for 40+ teams
Goose helps heritage organisations across the Fund’s network think through their work - now used by more than 40 of them.
Read the case study →Duchenne UK
Guidance families can listen to
Dense written guidance for families living with Duchenne, turned into audio they can take in on the move.
The Learning Lab
Knowing a clinician is ready
Simulated patients that test real clinical judgement, not whether someone can pass a multiple-choice test.
Read the case study →Your questions
The questions every team asks.
Caution is the right instinct with work like this. Each one already has a deliberate answer - built into how the software works, not bolted on afterwards to reassure you.
How it lands
Software you own, that we run.
Most of what you get is our standard software, well-tested across three years and many charities. The rest is connecting it to your own tools and data.
- A short discovery.We shape what the software should take on, with the team closest to the work.
- We build it.Standard software plus the integrations to your CRM, finance, monitoring and evaluation platform and other systems. Weeks, not months.
- We run it.It runs properly from the off, and each new task shares the same foundations - configuration, not new builds.
If a workflow is getting in your way, let’s have a chat.
Tell us the task, or grab a slot in the diary. Every message reaches us directly.