AI Leadership Principles / How to lead in the era of AI
AI Leadership Principles
Leading in the era of AI is less about the technology and more about decisions only leaders can make: which problems are worth solving, who owns the outcome, how people are brought along, and what must stay in human hands. These principles, and a practical guide to each question leaders ask most, are how I approach that work.
The principles
10 principles for leading with AI
Own it at the top
AI changes how the company makes money and spends its time, so it cannot sit with IT alone. The CEO and the executive team own the agenda, and one named person is accountable for delivery.
Start from the problem, not the tool
Begin with a delay, a cost or a customer need you can name. The technology comes second.
Redesign the work, not just the toolset
Handing out licences changes little. The value comes when a team rethinks the whole workflow: what the system does, what people do, and where a person decides.
Measure value, not activity
Logins and prompts tell you people are trying. Set a baseline, then track the business number the work was meant to move.
Lead through managers
People take their cue from the person they report to. I equip managers to lead adoption in their own teams, and I expect them to.
Invest in people as much as technology
Skills hold more companies back than models do. Give people time, practice on real work and someone to ask.
Learn it yourself
You cannot lead what you have not tried. Use AI every week on your own work, and say openly what you do not yet know.
Govern to go faster
Clear rules on data, decisions and risk let teams move without asking permission every time. Keep the guardrails proportionate to the stakes.
Be honest about jobs
Silence creates fear and hidden use. Say plainly that work will change, what you know, what you do not, and how you will support people.
Concentrate, then scale on evidence
Pick a few problems that matter and fund them properly. Scale what proves itself, and stop what does not.
The guides
The questions leaders ask, answered
Strategy
What should our AI strategy be, and where do we focus?
Start from your business strategy and the problems that matter most, not from the technology. Choose a few areas where the value is large, redesign them end to end, decide where AI should drive growth as well as efficiency, and be explicit about what you will stop doing.
Read the guide →Guide · StrategyHow fast should we move on AI, and how much should we invest?
Move fast on learning and adoption, and be disciplined about scaling spend. Fund in stages against evidence, concentrate investment on a few areas, and set realistic payback horizons. Fear of missing out is a poor reason to spend, and waiting for certainty is a poor reason not to.
Read the guide →Ownership
Who should own AI in our company?
The CEO and the executive team own the AI agenda, because it changes strategy, operations and people at once. Business leaders own the results in their own areas, a small central team owns shared platforms, standards and risk, and one named executive is accountable for delivery.
Read the guide →Guide · OwnershipShould we hire a Chief AI Officer?
Only if it fixes a real gap. If AI is central to your revenue or risk and nobody can make AI decisions stick, you need an accountable owner with a mandate and a budget. Strategic ownership still stays with the CEO and the executive team.
Read the guide →People
What happens to jobs and roles, and what do I tell people?
Be honest that the mix of work will change, and say what you know and what you do not. Pair any efficiency plan with reskilling and redeployment, redesign roles deliberately rather than by attrition alone, and give people a stake in the gains.
Read the guide →Guide · PeopleHow do we build AI skills across the organisation?
Build learning paths by role: everyone, managers, builders and risk owners. Make the practice hands-on, on people’s own work, with coaching. Protect time for it, redesign roles alongside the training, and keep a record of who has learned what.
Read the guide →Guide · PeopleHow do we lead teams where people work alongside AI agents?
Start with bounded, end-to-end workflows and name a human owner for every agent. Define what agents may do alone, what needs approval and when they hand over to a person. Redesign roles around supervising and improving agents, and measure outcomes per workflow.
Read the guide →Adoption
How do I get my people to actually use AI?
Treat it as a change in how people work, not a software rollout. Explain why it matters and what the plan is, equip managers to lead it in their teams, redesign a few real workflows, give hands-on training, and use AI visibly yourself.
Read the guide →Guide · AdoptionShould we mandate AI use or tie it to performance reviews?
A clear expectation can help, but only if it comes with tools, training and time. Assess the outcomes and quality of people’s work, not logins or usage counts. Let teams decide how they use AI, and share good examples openly.
Read the guide →Value
Why aren’t we seeing a return on AI, and how do we measure it?
Most companies measure activity, such as logins, pilots and projected savings, instead of the business result the work was meant to change. Set a baseline before you start, tie each initiative to one named profit, cost or customer measure, and review it at executive level every quarter.
Read the guide →Guide · ValueHow do we get from AI pilots to scale?
Design every pilot to scale from day one: a named business owner, a real workflow, a baseline and clear criteria for success. Redesign the work around it, build platforms and data you can reuse, and decide to scale or stop on evidence, not enthusiasm.
Read the guide →Risk
How do we govern AI risk and use AI responsibly?
Use a proportionate, risk-tiered approach: light rules for low-risk uses, stronger controls where decisions affect customers, money or safety. Set clear decision rights, monitor what runs in production, keep an incident process, and treat governance as what lets you move faster.
Read the guide →Guide · RiskShould we restrict ChatGPT, and how do we handle shadow AI?
Blanket bans tend to push AI use out of sight rather than stop it. Provide approved, secure tools quickly, write a short acceptable-use policy with clear data rules, train people on what not to share, and find out how AI is already used so the best ideas become official.
Read the guide →Technology
Is our data ready for AI, and what foundations do we need?
Probably not everywhere, and it does not need to be. Fix data one use case at a time, starting with the most valuable areas, give each important dataset a clear owner, and build on a shared architecture so tools do not multiply. Avoid multi-year clean-ups before any value.
Read the guide →Guide · TechnologyShould we build, buy or partner for AI?
Buy commodity capabilities, build only where your own data or process is what makes you different, and partner to borrow skills you cannot hire quickly. Judge every option on whether it fits your real workflow and learns from it, and keep a way out of every vendor.
Read the guide →Put it to work
From principles to your first decisions.
See what AI can do across your business, find the problems worth solving first, or get a tailored view of your company.
Leadership, not just tools
Need someone to lead
the AI agenda with you?
I work with a few leadership teams at a time as a fractional or interim AI leader: setting priorities, guiding delivery, and building capability that stays.