An AI strategic session for leadership teams who have bought the licences and not yet changed how the work gets done. Diagnosis first. Ideation last. Owners, metrics and a review cadence before anyone leaves the room.
Licences bought. Kick-off deck delivered. Months later the champion has quit, the tool sits unused, and someone is already pitching the next one doing much the same job.
Every stage on the planet has a keynote on how AI scales what exists. That includes broken processes - the ones your team already fixed with undocumented workarounds and their own spreadsheets. We then ask seasoned, capable people to accept that the way they have worked through an entire successful career is no longer good enough, and reward them by asking them to scale the broken version.
These three figures are the ones I open every keynote with, because they describe a market that has bought AI at scale and has not yet learned to run it. They are the reason a strategic session on AI is not a nice-to-have exercise in enthusiasm.
Nearly 90% of organisations now use AI in at least one business function, which is why the adoption question is no longer interesting on its own.
BCG
Despite that adoption, 60% of organisations report minimal returns. The licences arrived, the way the work gets done did not change.
BCG
Personal, hands-on governance of AI by the chief executive shows the strongest link to AI-driven EBIT of any factor measured. Not the stack, the calendar.
Thomson Reuters
Recurring conclusions from client sessions and industry panels, 2026
Most software rollouts follow the same plan. AI is not a new fancy CRM. It is a new way of working, an extra pair of brains alongside your colleagues. I think of a well-built AI skill as a highly professional team member - the one who just knows what to do and when, and performs consistently, almost.
That reframe changes the question. You stop asking where you can install AI, and start asking which outcomes could be better if the workflow behind them were redesigned.
You onboard a colleague. You give them context, boundaries, a review process and feedback. Licences do none of that, which is why they sit unused.
When delivery cost drops, the shape of what you can sell changes. Service tiers, response times and segments that were uneconomic become reachable.
Every leadership team has a dream service it never launched because it did not scale. Some of those are now affordable to deliver. Most companies never revisit the list.
The spreadsheet a senior manager built to survive a broken process is documented expertise. Captured properly, it is an asset. Left alone, it leaves with them.
Do not do AI for the sake of AI. Look for the value. Try doing something you have always dreamed of fixing or delegating, and experiment.
"We are too small to have strategic goals, and tracking them is too expensive." I hear that from portfolio companies before we decide how much to allocate to anything, AI included.
Strategy equals focus, and focus equals aligned effort. Fewer resources mean more focus is needed to get results, not less. It does not mean strategy is not worth trying. It means it is worth thinking about in advance. And strategy is not a one-man show, even when one person is running every function.
Every function owns its part of the goal, and that is what makes the whole goal achievable. Without that, an AI budget is a series of unconnected purchases.
You are not too small for strategic goals. You are doomed to stay small without them.
Not every organisation does, and I would rather say so on a landing page than three weeks into an engagement. Here is the honest sort.
This is the question the session exists to answer, and it is not answered by a tool comparison. Start with a friction analysis. Track the time spent on tasks. Ask team managers direct questions and detect bottlenecks before they solidify. Benchmark your own operational data against competitors, your industry, and industries that are further ahead. Fintechs hold the highest NPS across sectors. What can your customer support learn from them?
Then put every candidate workflow through four tests before anyone mentions a vendor.
Does the work consume time without creating proportional value? If nobody complains about it, it will not sustain the effort of redesign.
High volume and repetition are what make a redesign pay back. A quarterly task rarely justifies the build and the change management around it.
Is there data, and is it structured, historical and contextual? Bland inputs give bland outputs. That is why so many AI deployments look and feel the same.
If you cannot tolerate a mistake in this workflow, AI belongs beside it as a check, not inside it as the decision-maker.
These get collapsed into one in most readiness reports, which is how a company ends up technically ready and behaviourally unready, or enthusiastic and unable to integrate anything into real work.
| Question | What it examines | What a bad answer looks like |
|---|---|---|
| Can we adopt AI | Data, systems, security, integration, skills, governance, budget, decision rights | Data lives in scattered spreadsheets because my spreadsheet is better than your spreadsheet, and there is no single source of truth |
| Will people use it | Trust, perceived usefulness, psychological safety, incentives, role impact, manager behaviour | Executive enthusiasm reported upward, quiet abandonment reported nowhere |
| Will it create value | Baseline performance, workflow economics, customer and employee outcomes, quality, risk, scalability | Usage counted as value, with quality degradation and hidden labour unmeasured |
The sequence matters more than the content. Diagnosis before ideation, because otherwise senior participants advocate for projects they already wanted, technologies that are fashionable, and preferences their own department holds.
Team readiness, AI perception, resistance to change and business model review, benchmarked against 2026 practice. Interviews, workflow inventory, technology and data assessment, a risk scan and baseline KPI collection.
Employees, middle managers and senior leaders, on workflow readiness, attitude, trust and change capacity. Anonymised and reported by role and function. Used as a diagnostic instrument, never as a popularity poll.
Day one opens the frame and builds shared language. Day two establishes the honest starting point. Day three converts insight into choices with named owners, metrics and a 30-60-90 day plan. Full outline below.
A written summary carrying every insight, decision, portfolio choice and charter. Then twelve monthly notes and optional quarterly syncs, so the strategy does not go stale before its first review lands.
At 30 days, review whether owners acted and blockers were removed. At 60 days, review adoption, workflow performance, quality and risk events. At 90 days, decide whether to scale, redesign, pause or retire each initiative, with named actions attached.
A leadership team that can name its own AI portfolio, two or three workflows in production with owners and quality gates, a governance charter people actually use, and a documented decision on everything that was paused or retired.
Doing the homework is what makes the session productive. It sets the ground for informed discussion, data-based decisions and efficient use of the time in the room, and it produces a sense of ownership before day one. Every participant submits three things: one workflow they would protect, one they would redesign, and one they would stop.
Three days works because each one has a different cognitive job: exposure, then diagnosis, then divergence and convergence. Compressing it into one day produces enthusiasm and a list. I have run both.
A deep dive into case studies, AI strategy trends and benchmarks. Selected cases, sector evidence and the limits of each benchmark, so the group builds a shared reference frame and a common vocabulary for capability, economics, risk and competitive pressure. This is not motivational theatre, and external evidence is treated as a set of hypotheses rather than proof that you will get the same result.
Opportunity and threat mapping. What is changing, which assumptions are becoming unsafe, which customer, employee or competitor behaviours could shift, and which capabilities are mature enough to test now rather than speculate about. Output is a shared opportunity and threat map, not a strategy.
Current-state business model, customer journeys, value chain, core workflows, data flows, and a capability heatmap. Competitive position defined against what the industry is actually doing, not what it announces.
Constraints, boundaries and initial conditions - the starting point. Money, people, data, integration, time and decision rights. Risk appetite, governance, regulatory and brand boundaries. Baseline metrics, sources of advantage that must be protected, and the conditions that must be true for success.
Structured ideation around the priority workflows and the strategic questions surfaced on day two. Ideation comes last on purpose. Multiple framings are forced deliberately, so the group does not simply get better and better at describing the problem from one angle.
Options narrowed through explicit criteria: value, feasibility, adoption likelihood, risk, strategic fit and learning potential. Criteria built so they can point to one answer, then tested against one another. Three portfolios rather than one backlog - now, next, later. Each shortlisted initiative leaves with a named business owner, target workflow, baseline and target metric, dependencies, risk classification, human oversight model, pilot boundary, budget range and a first 30-day action.
The pack combines sources, and each source type answers a different question. Every statistic in it shows publication date, fieldwork date, geography, sample, respondent role, industry, question wording, and whether the measure is actual use, planned use, investment, expectation or outcome. No source, no claim.
| Source | Best use | Main caution |
|---|---|---|
| CB Insights | Market signals, startup activity, competitive moves, emerging agent categories, readiness patterns | Market signals are not proof of internal return. Definitions and methodology have to be checked |
| Statista | Survey statistics, adoption rates, sector and country comparisons | Many figures are secondary summaries. Record the original source, sample, date and question wording |
| PitchBook | Venture funding, private market activity, investor concentration, capital flows | Funding is a signal of expectations, not evidence of operational value |
| Academic research and Wiley | Adoption mechanisms, organisational readiness, resistance, strategy implementation, behaviour, governance | Peer review lags fast-moving technology, and samples can be narrow |
| Business journals and strategy experts | Strategic framing, cases, managerial heuristics, implementation narratives | Quotes and cases need context. Advice is not causal evidence |
| BMJ and NEJM Catalyst | High-risk implementation, workflow integration, human oversight, monitoring, accountability | Healthcare evidence is excellent on implementation discipline and does not transfer automatically to every sector |
I built this list with three colleagues from HR, operations and hiring, and have been correcting for it ever since. Every one of these is designed out of the flow above, which is the only reason the flow looks the way it does.
A strategic session is not team-building and not a corporate away day
The wrong people. The participant list does not match the task
The wrong time. The session is stretched too thin across the calendar
The wrong moment. It runs alongside annual budgeting and reporting
The CEO facilitates instead of participating, which costs the room their contribution and the session its objectivity
No homework. No financial analysis, benchmarking, market research, stakeholder interviews or process mapping done beforehand
No implementation plan for the results
No space for criticism, and no way to release the internal tension in the room
Just a conversation. No artefacts, and no agreement on what the artefacts should be
No control over implementation after everyone goes back to work
Zero onboarding for the team who did not attend but have to deliver it
No plan and no facilitation on the day itself
The design test: if the output cannot be turned into a budget request, a named owner, a workflow change, a KPI dashboard and a governance decision, it is not yet a strategy
CEOs tend to think that if they have shown something once, everyone grasped it. One-day training. A quick glimpse of the CEO using AI. A productivity statistic dropped into the corporate chat. In reality people need information repeated, and presented in different contexts, before they trust it. This is the biggest and most common gap I see.
So the session comes with a year of reinforcement, aimed squarely at the solutions and challenges the room actually defined.
One note a month for a year, written for your organisation rather than a mailing list. Relevant case studies, new research, and fresh data on the specific solutions and challenges defined in your session - the material that reopens the conversation at the right moment rather than a general AI newsletter.
A working session each quarter on deliverables and enrolment planning: what shipped, what stalled, who is blocked, and what the next cohort of people and workflows should be.
Quarterly syncs are scoped separately from the session and the monthly notes. If they are not contracted, the review cadence is handed to a named internal owner instead
Every perfectly planned boardroom strategy I have watched fail, failed because it did not account for how humans actually operate. Not because of a lack of investment. It is not primarily a change management problem. It is a psychological one.
You are asking senior, experienced, capable people in their thirties, forties, fifties and sixties to admit they need to start from scratch and step into a junior's shoes. That is an identity threat, and no amount of infrastructure investment fixes it.
Repeatedly rehearsing a problem strengthens problem-focused neural circuits and starves solution-focused ones of the repetition they need to become fast. Five days of leg day will not build your shoulders. Different sessions, different headspace, and "and then what" as the most useful two words in the room.
Method after Shane Parrish, Clear Thinking, 2023, applied with an early-stage quantum advisee choosing a business model
Losses feel roughly twice as painful as equivalent gains feel good, and people use far less information than they think to reach a final decision. Gathering more data feels like doing something. Deciding wrongly feels like your fault, while missing the window feels like the market's fault, which is exactly why managers sit on decisions for months.
Loss aversion: Kahneman and Tversky. Omission bias in decision avoidance
If one model sits at the core of a team, everyone produces the same ideas. In a study of people brainstorming with ChatGPT, 94% of shared ideas overlapped in concept and idea diversity dropped in 37 of 45 comparisons. Different people with different backgrounds should bring different ideas, so multiple cleaving frames are forced deliberately.
Wharton Mack Institute on the study published in Nature Human Behaviour, 2025
Co-created things are valued and trusted more than things that are simply bought. Engagement comes first: talk to the teams, get their perspective, workshop and ideate with them. Do not expect people to use AI because you said so. Engage teams in developing and onboarding the tool together, to make their life easier rather than your board report better.
In the largest qualitative study of its kind, the number one concern was not job loss. It was unreliability - AI that sounds confident and is confidently wrong. The number one use case was not automation either. It was professional excellence. That gap, between the excellence people want and the outputs they get, is where resistance lives.
Anthropic, study of 81,000 people across 159 countries, 2026
AI mimics behaviour beautifully and accurately detects patterns, and it does not understand context. An employee wellbeing system that flags late working as imbalance, without knowing about negotiated flexible hours and a newborn at home, harasses happy people and ignores struggling ones. The questions come first: what behaviour are we trying to change, how will we measure it, and what limits does that measurement have?
Behavioural AI framing after Professor Ganna Pogrebna
Surprise, then curiosity, then accountability, then persistence, then loop. A four-step sequence built to shorten the gap between "we bought the tool" and "people actually use it". It is why day one leads with evidence that unsettles the room rather than reassures it, and why the twelve-month follow-up exists at all - persistence is a stage, not a personality trait.
Facilitation is delivered with an ICF-recognised coaching credential and twenty years of running strategic sessions, including monthly sessions as a standing operating rhythm rather than an annual event.
An enterprise integration middleware platform for telecom and retail, in Cyprus and Ukraine. I joined as VP of Innovation and inherited the classic problem: strategic decisions took the better part of a year, and by the time one landed the market had moved.
We built a framework for fast market analysis and hypothesis testing, then ran monthly strategic sessions instead of one annual offsite. Decision-making time fell from nine months to three. I also launched a corporate university to close the gap between marketing, sales, account management and engineering - the same gap that stops AI rollouts today.
XME.digital, 2019-2025. Company internal figures. Named enterprise deployments: Ticinocom SA, Switzerland, a roughly $500K build over two years, and Vodafone Ukraine
The rhythm ran on four questions asked every month, connecting each result to the actions and the people that produced it. I wrote the method up while I was still running it, and it is the piece I most often send to leadership teams before a session.
Four questions to ask monthly to increase employee engagement and performance - my own write-up of the method, 2022. Ask me for the piece
on Perplexity for AI requirements tools, alongside IBM DOORS, at EltegraAI. 81 inbound enterprise booking requests, 35+ demos and 10 enterprise pilots including two NASDAQ-listed firms, with zero paid acquisition. AI-first go-to-market run by one person.
non-technical professionals trained as a Google AI Foundations trainer, across EMEA. More than twenty AI rollouts delivered, and more than thirty AI skills built by hand for sales, marketing, business development and strategy.
training engagement against a 70% benchmark, and 4.2 out of 5 for instructor quality, on the Exadel learning and development programme, 2023. Benchmark as reported by the client.
Fractional CMO, AI adoption adviser and deep tech go-to-market consultant. Twenty years across FMCG, automotive, telecom, healthcare, retail and deep tech, working from Limassol across EMEA. Board member. ICF-credentialed coach.
I use AI as a working tool and say so in every engagement. I am not a coder. Diagnosis before advice, and figures I cannot trace do not go in the room.
A facilitated decision-design process that turns an organisation's AI activity into a set of explicit choices. It runs in four stages: a readiness audit, an employee and manager survey, three facilitated days with the leadership team, and a written follow-up plus twelve months of reinforcement. It is not AI training and it is not a tool selection exercise. The output is a prioritised portfolio of initiatives with named owners, baseline and target metrics, governance decisions and a 30-60-90 day plan.
Three days in the room, preceded by two to four weeks of audit and survey work, and followed by a written summary. The twelve monthly notes run for a year after that, with optional quarterly syncs. The pre-work is not optional. A session without it produces enthusiasm, a slide deck and a list of use cases without changing operating behaviour.
Because each day has a different cognitive job. Day one expands the frame with external evidence. Day two establishes an honest starting point. Day three generates options and narrows them. Compress it and the group jumps to solutions, which means senior participants advocate for projects they already wanted before any shared diagnosis exists. One-day formats are possible and I will tell you what you are giving up.
The people who can approve budget, data access, policy exceptions and workflow changes, plus a technical lead who can connect ideas to systems, data, security and support capacity. Function heads whose workflows are in scope. The CEO participates rather than facilitates - a CEO in the facilitator seat costs the room their contribution and costs the session its objectivity. Frontline and middle-manager reality enters through the survey and the interviews, so executives cannot dominate the diagnosis.
Training builds capability in individuals. A strategic session makes organisational choices. Most companies need both, in that order, and buy them in the wrong order. Training a team to prompt well will make them better at running a broken process. Deciding which processes deserve redesign, and which must never be automated, is what the session is for. I deliver training separately, and the session usually specifies exactly which training is needed, for whom.
Fewer resources mean more focus is needed to get results, not less. Strategy is a decision-making tool for what you are and what you are not, and that tool matters most when you cannot afford to be wrong twice. Small does not mean informal: it means the session is shorter, the participant list is smaller, and the portfolio holds two initiatives rather than eight. You are not too small for strategic goals. You are doomed to stay small without them.
The short-term failure modes are consistent: the session begins with solutions, the audit is performative, executives dominate, the survey is not trusted, cases create hype without their transfer conditions, no decision rights are defined, and ideas are never quantified. Longer term: pilot purgatory, no workflow redesign, training as a one-time event, no measurement after launch, misaligned incentives, and leadership attention moving on. The design test is simple - if the output cannot be turned into a budget request, a named owner, a workflow change, a KPI dashboard and a governance decision, it is not yet a strategy.
Friction analysis first, then four tests on every candidate workflow: is it painful, is it frequent, is it structured enough to have usable data, and can it tolerate an error. If a workflow cannot tolerate a mistake, AI belongs beside it as a check rather than inside it as the decision-maker. The audit keeps three questions separate throughout: can the organisation adopt AI, will people actually use it, and will adoption create value. Companies get into trouble by collapsing those three into one.
A documented readiness baseline with a confidence rating. Survey findings segmented by role and function, with anonymity protected. A current-state business model and workflow map. A benchmark pack with source metadata and methodological caveats. A prioritised opportunity portfolio across now, next and later horizons. One-page charters for each priority initiative. Named executive and operational owners. Baseline and target KPIs covering adoption, quality, cycle time, cost, risk and customer or employee outcomes. A data, security, legal and governance checklist. A 30-60-90 day implementation plan. A review cadence with explicit scale, redesign, pause and retire decisions.
Baseline performance on the workflows in scope, whatever exists on data structure and systems, and access to the people who know how the work actually happens. Perfect data is not a prerequisite - most organisations still run on scattered spreadsheets, because my spreadsheet is better than your spreadsheet, and there is no single source of truth. Discovering that is itself a finding. What matters is that the audit produces an honest picture rather than a flattering one.
Yes, remote delivery is standard and on site is available by arrangement. Sessions run in English or Ukrainian. I am based in Limassol, Cyprus, and work across EMEA. Day three benefits most from being in one physical room, and if only one of the three days can be on site, that is the one to choose.
Price depends on organisation size, the number of functions in scope, and whether quarterly syncs are contracted. I quote after a 30-minute call, because the audit scope is what moves the number. EIC beneficiaries can access 50% co-financing via EIC ACCESS+.
If that is the honest answer, yes. Some organisations have one well-understood workflow and simply need it automated - buy the tool, run the pilot, measure it. Some have a strategy vacuum rather than an AI question, which is a strategic session without the AI prefix. And a session cannot work when the people holding decision rights cannot be in the room. I would rather say that on a call than three weeks into an engagement.
A 30-minute call, no deck. Come with one process you would protect, one you would redesign, and one you would stop, and we will know within half an hour whether a strategic session is the right instrument for you.
Limassol, Cyprus. Working across EMEA. English or Ukrainian