Strategic Intelligence.
Not Just Information.
Practical guides to AI governance, AI ROI, AI adoption and getting started with AI, written for UK business leaders running organisations of 15 to 200 people. ISO 42001, the EU AI Act, pilots that stall, costs that were not in the quote, and the human side that decides whether any of it lands. No hype. Just execution.
Twenty-eight guides across four pillars, plus a twenty-two term AI glossary. If you are not sure where to begin, start here.
Start here
Most organisations arrive at an AI library in one of three states, and the right first move is different in each. If you are not using AI yet, the work is finding one use case worth proving rather than writing a strategy. If AI is already in use but nothing governs it, the work is drawing a boundary around what is already happening before it becomes a problem you find out about from someone else. If AI is in use and governed but nobody can say what it returned, the work is measurement, and it is usually harder than the implementation was.
You are not using AI yet
The failure mode here is buying a platform before you have a use case, then trying to find work for it. Start from a process that is repetitive, high volume and low risk, and prove it on that before you commit to anything wider. The 45-minute route is the 3-5-4 Method; the longer answer is where a business should actually start. If budget is the constraint, the £69-a-month starter stack sets out what is achievable at the bottom end.
All getting-started guides →AI is in use, nothing governs it
This is the most common state and the least talked about. Staff have found tools that help and are using them, usually on personal accounts, often with company data. That is shadow AI, and the answer is almost never a ban, because bans move the behaviour rather than stopping it. Start with the three-layer governance model, then decide which standard you are working to with ISO 42001 versus the EU AI Act.
All governance guides →AI is in use, the return is unproven
Pilots that never end are the standard failure of the last three years, and the cause is usually that nobody agreed in advance what result would justify going further. Read pilot paralysis to production for the 90-day framework, the hidden costs of AI implementation for what the quote did not include, and the real AI ROI timeline for what a defensible number looks like and when it arrives.
All ROI and delivery guides →The four questions this library answers
Everything below is written from UK engagements with organisations of roughly 15 to 200 people. Where a number appears, the article it links to shows where it came from.
What is AI governance, and what does a UK SME actually need?
AI governance is the set of decisions an organisation makes about what AI may do, on whose data, with whose approval, and what happens when it gets something wrong. For a UK business under 200 people it does not need to be a management system with a hundred controls. It needs three things written down: which data may go near a model and which may not, which decisions a person must sign before they take effect, and who owns the answer when something goes wrong. Most of the value is in having decided, not in the length of the document.
The two frameworks people ask about are ISO/IEC 42001 and the EU AI Act, and they answer different questions. ISO 42001 is a voluntary management-system standard: it describes how you run AI responsibly, and you can build to it without certifying against it. The EU AI Act is law, it classifies systems by risk, and it reaches UK businesses whose systems affect people in the EU regardless of where the company sits. A UK SME selling only domestically and using AI for internal drafting and analysis is in a very different position from one whose model screens job applicants or affects access to credit.
The practical order is usually: find out what is already being used, decide the data boundary, write the policy that reflects the decision rather than an aspiration, then pick a framework to align to if a customer, insurer or investor is going to ask. Doing it in that order takes weeks. Doing it in reverse takes quarters and produces a document nobody follows.
Read next: the three-layer model · what the EU AI Act means for UK SMEs · where your AI data is actually stored · how we deliver governance
What is a realistic return on AI, and how long does it take?
A realistic return on AI in a small or mid-sized business comes from hours given back on repetitive work, not from headcount removed, and it shows up in months rather than weeks. The pattern we see is that the first measurable result arrives around 90 days after a use case goes into real work, and that it is almost always smaller and more boring than the business case predicted, because the business case was written about the best day rather than the average one. The organisations that get a defensible number are the ones that measured the process before they changed it.
The most common reason a return cannot be proven is not that the AI failed. It is that nobody recorded how long the task took beforehand, so there is no baseline to compare against. The second most common is that the cost side was only ever counted as licences. Licences are usually the smallest line. The real cost is the time spent configuring, the time spent training, the time the process spends running twice while people build confidence in the new version, and the ongoing maintenance nobody assigned to anyone.
The fix is to agree, before anything is built, what result would justify going further and what result would mean stopping. Both numbers, written down, with a date. That single discipline separates a pilot from pilot purgatory more reliably than any technology choice.
Read next: the real AI ROI timeline · the hidden costs of implementation · dark data, and why projects stall on it · the 90-day framework
Why do AI adoption programmes stall, and what fixes it?
AI adoption programmes stall for human reasons far more often than technical ones. The tool works; people do not use it. Underneath that, the usual causes are that staff quietly believe the tool is there to replace them, that nobody senior is visibly using it, that the training was a one-hour demonstration rather than practice on real work, and that using it properly is slower than the old way for the first fortnight and nobody warned anyone that this would happen.
The fix is unglamorous. Say out loud what the organisation intends to do about roles, because the absence of a statement is heard as the worst possible answer. Train on the work people actually have, not on a generic prompt exercise. Give the first fortnight explicit permission to be slower. And put the capability where the work already is, rather than asking people to visit a separate tool and remember to. Adoption is a culture problem wearing a technology costume, which is why we treat it as one.
There is a measurable version of this. The gap between what an organisation could do with AI and what it does do is mostly capability, not licences, and it can be mapped person by person rather than guessed at department level. That mapping is what tells you whether the next pound is better spent on tooling or on teaching.
Read next: the human capability equation · what staff are actually worried about · six UK SMEs who upskilled instead · how we deliver training
How should a UK SME with no AI in place start?
Start with one process, not a strategy. Pick something repetitive, high volume, low risk and currently done by a person who would rather be doing something else: drafting standard documents, summarising long inputs, reconciling information that lives in two systems. Time it as it is today. Run it with AI for a fortnight alongside the existing way. Compare. That gives you a real number, a team that has now used the thing, and a decision you can defend, all before you have committed to a platform.
What not to do first: do not start with a tool comparison, because the answer depends on the use case you have not chosen yet. Do not start with a company-wide rollout, because you will be training people on something you have not yet proven. And do not start with a policy document written in the abstract, because a policy written before anyone has used AI governs an imaginary organisation.
On platform choice, most organisations of this size end up running more than one, and the useful question is which job each is for rather than which is better. Claude versus Copilot sets that out without a feature matrix. And if the question behind the question is whether any of this needs custom software, it usually does not: bespoke AI for a business this size normally means configuring an existing platform around how you already work.
Read next: where to start · the Monday morning method · the tools comparison · the free readiness assessment
The vocabulary, in plain English
A lot of AI conversations stall because two people are using the same word differently. These are the terms that come up most often in UK board discussions, each defined so it stands alone. The full glossary has twenty-two.
The Ignite Almanack
One idea, made useful by Monday. Short, practical guides for UK business owners putting AI to work, each one taking a single thing we do for SMEs and making it usable this week.
Real numbers, copy-paste prompts, plain English. They come from actual client work, not theory. Download any of them and share them with your team. There is no email wall and nothing to sign up for.
Field Guide 01
AI Leverage
How a small business multiplies its output without multiplying its headcount. The third form of leverage, after Naval Ravikant, and the one that needs nobody’s permission.
Field Guide 02
Claude or ChatGPT
An honest 2026 buyer’s guide for the UK business choosing between the two. At team tier they cost about the same, so the seat price is the wrong thing to argue about.
Field Guide 03
Shadow AI
The AI your team already uses, and how to bring it into the light. Shadow AI is not a discipline problem. It is a map of your real pain with a UK GDPR exposure attached.
Field Guide 04
Governance in a Week
How to make AI safe in a week, so your team actually uses it. Write the red lines first and adoption follows. The human manifesto and the ISO 42001 starter, side by side.
Field Guide 05
Knowledge Capture
Getting what your best people know out of their heads and onto the page. The 9.3 hours a week lost hunting for answers, and the knowledge that walks out with a resignation.
Field Guide 06
The AI Operating System
Turning AI from a thing that answers questions into a system that does the work. Business context, your voice, role skills, and a knowledge base that compounds.
Field Guide 07
Why AI Pilots Fail
Pilots stall because nobody agreed what the problem was. The MD sees one problem, ops sees another, finance sees a third. One page, written first, kills the bad ones cheaply.
Field Guide 08
The AI Ladder
Eight rungs, two questions, and the one move worth making this month. Every task sits on its own rung, and most firms put all of theirs on the same one.
Field Guide 09
The 90-Day Proof
How to know your AI programme is working before you have paid for all of it. The gate at day 90, the baseline you have to take on day zero, and what to measure.
A new field guide lands every few weeks. Tell us what to write next, or go past one idea with a SPARK Discovery.
Research and analysis
The Lead Response Crisis
£104 Billion Annual Revenue Loss. Our analysis reveals 73% of leads are lost due to a 47-hour average response time.
The Human Capability Equation
Why most AI projects stall. The cause is the multiplication-by-zero of leadership, culture and governance, rarely the technology.
Where Should Your Business Start?
A process-first guide to finding your highest-impact AI opportunity in 90 minutes, not 90 days. Stop pilot paralysis.
The 2026 UK SME Guide to AI Governance
From Manifesto to ISO 42001. How to control AI without slowing innovation. The culture and compliance roadmap for UK SMEs.
The Hidden Costs of AI Implementation
Software licences are only 30-50% of total AI costs. Here's the complete cost breakdown UK SMEs don't see coming.
£3.70 Return Per Pound: The Real AI ROI Timeline
Years 2-3 cost £40k-70k annually. Here's when AI actually pays off and what the 30% who succeed do differently.
AI on a £69/Month Budget: The Starter Stack
44% of UK SMEs use off-the-shelf tools successfully. Here's the practical £69/month AI stack with 952% Year 1 ROI.
Will AI Take My Job? What UK Employment Data Shows
83% of UK workers say AI enhances creativity. High-AI sectors show 2.3% employment growth, not decline. Here's the data.
AI Without Redundancies: 6 UK SMEs Who Upskilled Instead
89% retention when upskilling vs 34% without. Real case studies across retail, legal, services, HR, manufacturing, finance.
Dark Data: The Silent Killer of Your AI Project
40% of UK AI projects fail due to fragmented data. Here's the 72-hour audit framework to identify dark data before you invest.
The £60bn Question: Where Is Your AI Data Actually Stored?
73% don't know where AI data is stored. Cloud sovereignty, GDPR compliance, and the vendor audit framework you need.
From Pilot Paralysis to Production: The 90-Day Framework
70% of UK AI pilots never reach production. Here's the 90-day deployment framework used by 40+ UK SMEs.
The August 2026 Deadline: EU AI Act for UK SMEs
Despite Brexit, UK businesses trading with EU face compliance. Penalties up to €35M. Here's the 6-month roadmap.
ISO 42001 vs EU AI Act: Which Standard for UK SMEs?
Decision framework: voluntary ISO certification vs mandatory EU Act compliance. When you need one, both, or neither.
Legal Sector AI: 75% Less Report Writing Time
SRA-compliant AI implementation for UK law firms. Contract review, legal research, client correspondence automation.
Retail AI Beyond ChatGPT: The E-commerce Stack
Complete 6-layer AI stack for UK retailers. 94% faster product listing, 31% basket recovery. £59k value created.
Professional Services AI: Reclaiming 10 Hours/Week
Time savings for consultancies, accounting, advisory services. 28% revenue per employee increase with quality frameworks.
Don't Miss the Next Insight
Join 2,000+ UK leaders receiving our strategic intelligence.
