An AI charter for an interiors brand.
House of Hackney · audit, charter, approved tools and training
House of Hackney now has one answer on which work AI may carry and which stays with the makers. It was agreed before the tools were bought. We wrote it from what all ten departments told us about the AI they were already using.
Overview
House of Hackney is a B Corp luxury interiors company of around 45 people across London and the US, and what it sells is work done by hand. AI had already arrived inside the business, in inboxes and briefs and whatever tools individual people had picked up, but it had arrived informally, with nothing agreed and nothing written down.
They brought us in to settle it properly, in terms the whole company could live with. The audit came first, the charter was written from what it found, and the approved-tools register, the training, the onboarding pack and a standing governance forum were all built on top of it.
- Every department named the parts of the work it considered untouchable.
- The charter is written in the company’s own language, short enough that people quote it to each other.
- It was adopted across all ten departments, with the Managing Director accountable for it at board level.
35
people surveyed across all ten departments, followed by a deep-dive session with each one.
57%
were already using AI at least a few times a week, with no shared policy sitting behind any of it.
0
daily users rated themselves very confident, which is the gap the training had to close.
83%
were open to training; what was missing was an agreed line.
20 hrs a week
back in finance, once the reconciliation work was automated.
Challenge
What the charter had to settle
The question coming out of every department was the same one: what are we actually allowed to do with this? For a company whose value sits in work that a person can point at and say they made, that question runs straight into the brand, into the environmental cost of the tools, into what happens to people’s jobs, and into whose data ends up on somebody else’s server. The charter had to answer those questions in writing:
- What is everyone allowed to use, and with what kinds of data?
- Which work stays with people, in writing?
- Who keeps oversight when the tools change again?
Approach
From audit to adopted charter
Survey and interviews across all ten departments
A survey of 35 people, then a deep-dive session with each department, asking where AI was already in use and where it worried people.
A charter in four layers
The charter runs from what the company believes, through what it commits to, down to how it decides and which tools it actually uses. It is specific enough to settle an argument about a particular piece of work.
Approved tools, with data rules
A maintained register of what everyone can and cannot use, and with what kinds of data, alongside an assessment record so that a new tool has to earn its place before anyone puts it to work.
A managed Claude environment
Claude set up for the house and properly run, configured against the charter’s data rules, with someone inside the company accountable for how it is used.
Training and an onboarding pack for new joiners
Skills sessions across the teams, then a pack that tells a new starter what AI is and how this house uses it, plus custom automations built for the specific jobs each department kept losing time to.
An AI governance forum with board accountability
The Managing Director chairs the forum. It holds the document and takes the decisions that keep it current.
The line
What the charter allows and rules out
The charter clears AI for routine administrative work and for first drafts that a person then rewrites, and it asks anyone who used AI on a piece of work to say so. It keeps the creative work and the house’s own voice with people, and customer data goes only into tools the charter has cleared.
The signed document stays private to the company.
Result
What changed
The company now works to one written position on AI, down to which tools are approved and what data those tools may see. The AI Governance Forum owns it from here.
- New joiners meet the onboarding pack before they meet the tools, so nobody has to work out the boundaries by guessing.
- A new tool now goes through the assessment record to get in.
- OpenKit automated reconciliation in the finance team once the rules were settled, returning twenty hours a week to the department.
Other engagements
What could your business do with AI?
Find out with experts who become part of your team. We uncover opportunities, get ideas working, and help your people build on the results. We reply within one working day.
Get in touch Governance & compliance