Governing Judgement
The guide to adopting AI in the built environment without losing professional judgement or accountability.
AI changes who drafts, not who decides. Governing judgement means a competent, named person still decides, and can show it.
construction professionals in a global RICS survey use AI in some part of their work. Less than 1% say it is fully integrated across their organisation.
Use is individual. Organisations have not caught up.
of UK businesses using AI have a formal written AI policy. A further 12% have informal policy or guidance.
Most firms using AI have no written rules.
of architects' practices use AI on at least some projects. Around 19% have a documented AI policy.
Policy lags use in design practice too.
of architecture, engineering and building employees in the UK and US have used AI tools their employer had not approved.
Shadow AI is already on your projects.
of cited sources supported the claim made, across the strongest of 14 AI systems tested, although over 94% of the links worked.
A working link is not a checked source.
of agentic AI projects are expected to be cancelled by the end of 2027, citing cost, unclear value and weak risk controls.
Agents need a business case and boundaries.
Six stages. 23 tools. One page each.
Every tool has a graphic, steps, a prompt, a worked example and a free Word or Excel template.
Decide
Set the rules before the work.
Use
Keep the judgement with the person.
Check
Check the claim, not the link.
Prove
If you cannot show it, you did not check it.
Review
Governance that does not move is paperwork.
Agents
Decide what it may do before it acts.
Four roles. Five tools each. Start with the first.
Practitioner
You use AI in your own work to produce drafts, estimates, summaries, calculations and reports.
- Data Boundaries
- Team Prompt Standard
- Show the Workings
- Claim Audit
- Judgement Self-Check
Reviewer
You approve AI-assisted work by others, and your name or the firm's goes on it.
- Checking Level Matrix
- Claim Audit
- Evaluator Red Team
- Use and Decision Register
- Agent Boundary Card
Steward
You own the policy, the registers and the evidence an assurer would ask for.
- AI Policy on a Page
- AI Tool Card
- Evidence Library
- Near-Miss Log
- Quarterly Governance Review
Sponsor
You hold executive accountability and answer when a regulator, insurer or client asks who decided.
- AI Policy on a Page
- Supplier AI Disclosure Reply
- Quarterly Governance Review
- Board Agent Questions
- Agent Maturity Check
Every tool works the same way.
Copy it, adapt it, fill it in and argue over it in your own team.
At a glance
What it is for, who uses it, how long it takes and what it links to.
Why it matters
The rule, then the evidence behind it, with numbered sources.
How to use it
Three or four steps, plus a prompt that keeps the judgement with you.
Worked example
The template filled in, so you can see what good looks like.
Template
An editable Word or Excel version of every tool.
If one tool changes how one decision is made on one project, it has done its job.
Why we are giving this away
Feenix wants the built environment to adopt AI responsibly, because done well it will raise standards across the industry and improve outcomes for the people we design, build and maintain for.
We believe AI can help close the skills gap, but only if we train people early: before habits form, and while the tools are new to almost everyone. Left to chance, AI will widen the gap between people who know how to check it and people who do not.
This handbook is a valuable asset, and we have made it free on purpose. Its real value is not in these pages. It is in the people who make up the built environment recognising the opportunity and the challenge in front of them, and deciding to meet both.
Get your free copy of Governing Judgement
Read it in your browser or download the PDF. We will also email you a copy so it is there when you need it.