Insights
Governing AI, in practice.
Plain, useful writing for boards and leadership teams. Filter by topic, industry or format.
Free reports
AI Governance for UK Boards
A practical primer for directors: why AI is now a board issue, the questions to ask, and the frameworks you can align to.
Download PDF
Responsible AI in Practice
How governance is engineered into the code: the six controls we build into AI systems, and what they look like in production.
Download PDF
AI in Regulated UK Sectors
A field guide for financial services, the public sector and the professions: the rules that already bind your AI today.
Download PDF
Why AI Projects Fail: The Evidence
The failure statistics boards actually get quoted, with what each one really measured, the six failure modes behind them, and the questions that prevent a repeat.
Download PDF28 results
AI governance for charity trustees: what changed
The 2025 Charity Governance Code now names a technology and AI policy as evidence of good governance. What trustees must do, sized for small charities.
The AI governance framework UK organisations actually need
A working AI governance framework has five connected layers — principles, policy, controls, evidence, assurance — and a 90-day route to stand one up.
AI governance for housing association boards
Repairs triage, arrears scoring and complaint handling are going algorithmic. What housing association boards must evidence — and to whom.
AI governance for school governors and MAT boards
What school governors and MAT trustees actually oversee on AI: DfE expectations, safeguarding, assessment integrity and the questions to ask.
The AI-native consultancy: consulting without the pyramid
The consulting pyramid amortised the cost of analysis. AI removes that cost — what AI-native honestly means and the proof clients should demand.
What a UK AI policy must include in 2026
The eight working parts of a defensible UK AI policy, what each section is for, and why a template without controls is a disclaimer, not governance.
Does the EU AI Act apply to UK organisations?
Three routes pull UK organisations into the EU AI Act. A plain-English decision guide for boards: scope, roles, risk tiers and the June 2026 timeline.
ISO 42001 vs NIST AI RMF: which do you need?
One is a certifiable management system standard, the other a voluntary risk framework. How a UK board chooses between them — or runs both inside one AIMS.
20 questions every UK board should ask about AI
Twenty AI questions for UK boards, grouped into five areas, each with the artefact a good answer produces and the UK rule it rests on.
Shadow AI: the policy boards need before the ban reflex
Staff already paste work into consumer AI. The answer is not a ban: discover use, triage it into three bands, provide sanctioned tools, police the line.
Top AI consultancies in the UK (2026): a buyer's guide
Ten UK AI consultancies compared on regulatory fluency, build capability and board-level focus. Every firm verified — and one of them is ours.
Why AI projects fail: what the numbers actually say
80%, 95%, 42% — the famous AI failure statistics measure different things. What each number actually says, what it leaves out, and the gap they all point to.
Read-only by construction: AI that cannot change what it reads
How a Postgres read-only transaction, a confidence gate and a cost-approval gate stop an analytics AI from writing to data it should only read.
Confidence floors and reason codes: when code overrules the model
How a configurable confidence floor and enumerated query codes keep an AI out of the ERP — and why deterministic code, not the model, decides to post.
Make every AI claim quote a real source, or fail
How we force every AI quotation to be a literal substring of the source, and block a denied vocabulary in code, in our insolvency build.
What a combined authority asks for before AI goes live
The pre-go-live checklist a UK combined authority demands of an AI tool: ATRS record, DPIA and Article 30, NCSC's 14 principles, advisory-only design and chunk-level citations.
The append-only decision ledger AI governance needs
How an append-only, no-update-no-delete ledger of named human accept/modify/reject decisions becomes the audit trail and RICS K5 disclosure regulated AI work requires.
The pacing problem: capability outpaces board ratification
AI capability changes between meetings, but boards govern on an annual cadence. The fix is standing governance engineered into the system, not a yearly sign-off.
What a RICS AI disclosure teaches every regulated profession
A surveying build generated its AI disclosure from real decision records. Here is how to turn that into a disclosure template for any regulated profession.
Make your AI risk register living evidence, not a spreadsheet
An AI risk register that only updates quarterly is already stale. Structure it with NIST's Govern-Map-Measure-Manage and feed it from the systems themselves.
No AI rulebook, but your AI is already bound
There is no FCA AI rulebook. Consumer Duty, SM&CR, SS1/23 and UK GDPR's new Articles 22A-22D already govern AI in financial services today.
ISO/IEC 42001 explained: what it asks of a board
What ISO/IEC 42001 concretely requires of a board across clauses 4-10 and Annex A, and the honest difference between aligning to the standard and being certified.
Governing the Intelligence Age: capability you did not build
The defining governance problem of the Intelligence Age is accountability for AI capability your board buys rather than builds. Here is what that asks of you.
The UK has no single AI Act. What your board governs instead
There is no UK AI statute. Your board governs against five voluntary, regulator-applied principles, which makes voluntary frameworks the practical route to compliance.
AI Governance for UK Boards
A practical primer for directors: why AI is now a board issue, the questions to ask, and the frameworks you can align to.
Responsible AI in Practice
How governance is engineered into the code: the six controls we build into AI systems, and what they look like in production.
AI in Regulated UK Sectors
A field guide for financial services, the public sector and the professions: the rules that already bind your AI today.
Why AI Projects Fail: The Evidence
The failure statistics boards actually get quoted, with what each one really measured, the six failure modes behind them, and the questions that prevent a repeat.
Find out where your AI exposure sits.
We'll tell you plainly what's worth doing, what isn't, and what a board or regulator will expect to see. No pitch deck.
No obligation · no pitch.