A direct, confidential first conversation. No presentation required.
We design value-led IT and AI strategies, sustainable architectures and governed adoption paths—from use case to scale.
The decisions we address
We clarify the real decision, the value at stake, constraints and viable alternatives. Assumptions and trade-offs become visible before resources are committed.
From evidence to roadmap
We combine stakeholder evidence, economics, capability assessment and risk. The outcome is a prioritised route with owners, dependencies and measurable success criteria.
Execution built into the design
Governance, technology, people and change are designed together. We can stay through mobilisation and delivery, then transfer methods and capability to the organisation.
Delivery
How the engagement works
01 · Evidence
We reconstruct facts, constraints, stakeholders and open decisions.
02 · Options
We compare alternatives, impact, risk and required capabilities.
03 · Mobilisation
We define owners, roadmap, governance and first executable decisions.
04 · Transfer
Methods and tools become part of day-to-day work.
What stays with the client
A documented decision, a governable roadmap, clear accountability and the assets required to continue without dependency on the adviser.
Metodo
How we work on this
Durations are indicative and shift with the size of the organisation. What does not shift is what goes into each phase and what comes out.
01
Current-state picture
2–3 weeks
In: the architecture as documented. Out: the architecture as it actually is, with technical debt quantified in days and in risk rather than described with adjectives.
02
Use case assessment
2–4 weeks
In: data and AI initiatives running or proposed. Out: which ones survive a value calculation, which are experiments and should be called experiments, which should be stopped.
03
Governance and target architecture
3–5 weeks
In: selected use cases, regulatory constraints, available data. Out: target architecture, a governance model for data and models, control points against the AI Act and ISO/IEC 42001.
04
Production and oversight
variable
In: the approved architecture. Out: the first capability in operation with monitoring, assigned roles, and criteria for switching it off if it misbehaves.
Deliverable
What the client is left with
AI use case register
Each case with expected value, required data, risk, AI Act risk tier, and a decision: proceed, experiment or stop.
Target architecture
Diagram plus decision notes: why that component and not the alternative, what breaks if it changes, what it costs to reverse.
Governance model
Who approves a model, who monitors it, which thresholds trigger a review, what gets documented and for how long it is kept.
Technical debt paydown plan
Interventions ordered by risk and by the cost of postponing them, not by the technical preference of whoever proposed them.
Decisioni
The decisions brought to us
These are the questions as the decision-maker puts them, not as we would rephrase them.
Does this AI pilot go to production or do we stop it — and by what criterion, decided before we grow attached to it?
Is our data ready for the use we have in mind, or are we building on a base that will not hold?
What has to be in order before the AI Act becomes a compliance problem rather than a roadmap item?
How much technical debt can we still afford to defer, and what happens if we defer it another year?
Misura
How we tell whether it worked
Indicators are agreed before we start and reviewed together. None of these is a promise of a result: they are how a result gets measured.
Use cases stopped deliberately
How many were closed by an explicit decision instead of being left to die when the budget ran out.
Time from data to decision
How long between a data point becoming available and its use in an operational choice.
Governance coverage
Share of models in production with an owner, alert thresholds and current documentation.
Cases
Cases related to this service
Refresh the hardware or change the model: a choice deferred too many times
Context
Mid-market industrial group. Production systems were slowing down on a physical infrastructure at end of life.
Constraint
Stopping production was not an option. And replacing the hardware would have brought the same problem back within a few years.
Decision
Migration to public cloud instead of a hardware refresh, with the change windows placed outside production hours.
Outcome
Not a single minute of downtime during working hours. Annual operating and maintenance costs down 32%.
Case anonymised with the client's authorisation. Metrics refer to the perimeter described; baseline and period are shared in a confidential conversation.
FAQ
Frequently asked questions
How does an engagement begin?
With a direct conversation and, when useful, a short assessment to clarify the decision, evidence, scope and expected outcome.
What do you deliver?
A decision-ready view and the assets needed to act: options, business case, roadmap, governance, measures and delivery plan as relevant.
Do you support implementation?
Yes. We can lead or assure delivery and transfer capability to internal teams.