Ask Your Data / for charities and support services

Ask your charity's records a question. Get the answer in seconds — kept in the UK.

Staff type a question in plain English. The answer comes from your own case records, with personal details hidden and a person checking anything before it goes out.

  • UK-hosted
  • Personal details masked
  • Every question logged
  • A person approves
85-second demo · captions on screen · all data shown is fictional

Before and after

From three days to one question.

Before

Three days in spreadsheets.

The funder report is due. Someone exports the case list, filters it by hand, cross-checks outcomes against notes, and copies figures into a document, hoping nothing was missed.

  • Exports from several places
  • Manual filtering and counting
  • No record of how the figures were reached

After

outcomes for housing cases this quarter

A checked draft in about a minute.

The figures come back with the records they came from. A draft paragraph is ready, and a person reads it and approves it before it goes anywhere.

  • Answers traced to your own records
  • Personal details hidden throughout
  • Approved by a person, and logged

Safe by design

Your records stay yours.

Built for the data-protection questions trustees and managers rightly ask. In plain words, this is what it does and doesn't do.

It only sees what you allow
You choose which systems, fields and records it can read. Everything else is out of reach.
Personal details are masked
Details such as names, addresses and dates of birth are hidden before the AI model sees anything.
Every question is logged
Who asked what, when, and which records were read. You can look back at any answer.
A person approves
Nothing goes out, whether a report paragraph or an email, until someone on your team has checked it.
Data stays in the UK
UK-hosted from end to end, and tested before go-live so you can see where each request goes.

What staff could ask

Questions you already ask, answered faster.

It works alongside Seren, or the case system and spreadsheets you already use. Nothing to migrate, and no new system for staff to learn.

  • OutcomesWhat were the outcomes for housing cases we closed this quarter?
  • CaseloadsHow many debt cases are open, by caseworker?
  • FundersPull together the figures for our quarterly funder report.
  • Waiting timesWhat's the average wait from referral to first appointment?

How it works, for your IT team

A governed connector between your systems and the model.

At its centre is a custom MCP server. MCP, the Model Context Protocol, is the open standard for connecting AI to tools and data.

  1. 01 · your systems Case records, spreadsheets, documents stay where they are
  2. 02 · MCP server Permissions, masking, audit log UK-hosted gatekeeper
  3. 03 · model The model of your choice UK-hosted or self-hosted open model
  4. 04 · approval A person checks the answer before anything goes out
  • Connector

    A custom MCP server sits between your systems and the model. The model only reaches data through it.

  • Model

    The model of your choice: a UK-hosted model or a self-hosted open model on servers you control.

  • Access

    Field-level permissions and role-based access, so each person's questions only reach what their role allows.

  • Masking

    Personal data is masked in the connector, before any request reaches the model.

  • Audit

    A full audit log of every question, every record read and every approval.

  • Approval

    Human approval gates on anything that leaves the system.

  • Testing

    Accuracy and error tests against your own questions before go-live, plus checks on where requests go.

  • No lock-in

    MCP is an open standard, so you can change model later without rebuilding the connection.

Questions

Straight answers.

If yours isn't here, ask us. We'll tell you plainly if this isn't the right fit.

Does our data leave the UK?

It's designed not to. The connector runs on UK hosting and the model is either UK-hosted or an open model run on servers you control. We agree the exact arrangement with you, and test where every request goes, before anything goes live.

Do staff need training?

Very little. Staff ask questions in plain English. We walk the team through it, including how to check an answer and how the approval step works.

Does it replace our case system?

No. It works alongside the system you already use, whether that's Seren, another case system or spreadsheets, and only reads what you've allowed.

Which AI does it use?

The model that fits your organisation: a UK-hosted model or an open model run on servers you control. Because the connection is built on MCP, an open standard, you can change model later without rebuilding it.

Can it change or send anything on its own?

No. It reads what you've allowed. Anything that would go out, such as a report paragraph or an email, waits for a person to approve it.

Want to see what your own records could tell you?