An open community for public sector data and AI

Build once.Use everywhere.

An independent, open-source community for the people who build public sector data and AI, fixing in the open the problems none of us can fix alone.

The problem

The problem we all know

A team solves a hard data problem. Two years later another team, often in the next building, pays to solve it again. Data sits in platforms that are hard to leave, and AI bills climb faster than anyone planned.

Why now

The builders need to be in the room

The next two years will shape public sector AI for a decade: who controls public data, what it costs to move it, and whether a tool built by one council reaches the next.

How we work

Borrowed from the open-source communities that last.

01

Community before code

We grow builders and brief decision makers. Good software follows.

02

You join as yourself

Not as your employer. Every member has the same voice.

03

Those who do, decide

Influence is earned by contribution, not job title or sponsorship.

04

Open by default

If it isn't written down and shared, it didn't happen.

05

Nobody sells

Suppliers and sponsors join on the same terms as everyone else.

06

It only counts if it's used

Delivered means running in a public sector organisation.

What we will do

Phase one runs from November 2026 to March 2027.

Bring us a decision

Before you commit to an architecture or a supplier, get an independent red team from peers who have done it before.

Build the shared foundations

A member-designed portability test, open reference architectures, and labs with synthetic and open data.

Ship things that get used

Open-source tools for real public sector problems, each with a named owner who will use them.

Make sure builders are heard

Members present their work to senior leaders and Parliament. We are always clear about who funds us.

The first sessions

Monthly, hybrid, about two hours, built around a real problem and a live demo.

  1. 25 NOVFounding session. Shaping Groundwork and agreeing our charter, with a live demo of Afon
  2. DECSynthetic data: targets, when it is good enough, and how to make it
  3. JANPublic launch. Why open source matters for public services, with Nesta
  4. JANAI cost control: getting value from every token
  5. FEBThe portability test: designing it together
  6. MAR[NEET data with OneLondon, to confirm] and plans for phase two

Advisory board

Senior advisers who guide Groundwork and keep it independent. More members will be announced soon.

Dr Laura Gilbert CBE

Dr Laura Gilbert CBE

Founding director of the No 10 data science team and the Incubator for AI
Chair
Polly Jones

Polly Jones

Co-founder, Emrys Health; former Head of Delivery and Products, NHS England
Mallory Durran

Mallory Durran

Group Head of Data Science, Nesta
Coming soon

To be announced

Coming soon

To be announced

Coming soon

To be announced

Coming soon

To be announced

Coming soon

To be announced

Coming soon

To be announced

Who is behind Groundwork

Groundwork is independently chaired by Dr Laura Gilbert CBE, guided by a senior advisory board, and convened and administered by Emrys Health.

Phase one is funded by Snowflake through a contract with Emrys Health; Emrys co-founder Polly Jones also works part-time for Snowflake. Sponsors fund running costs only. They do not set the agenda, approve outputs or get access to members for sales. We welcome further sponsors and will publish every funder.

Bring a real problem.

We are looking for up to 40 founding members. Email us by Monday 23 November with your role and the problem you want to solve.

info@groundworkuk.tech

Pass this to someone who builds. Groundwork will only ever be as good as the people in it.