AI strategy starts with three questions.

You already know AI exists. The useful conversation is narrower: what it changes in your company, in which order, and who owns the result.

Product
Production
Organisation

The framework

Three questions every management team should answer.

Product · Production · Organisation

Most companies begin by asking where AI could make them more efficient. That is a useful question. It is also the smallest of the three.

Product

AI in what we sell

How does AI change our product, our service or our business model?

Growth & Differentiation

Revenue
Customer value
Differentiation
Business model

Can AI improve our existing product?
Can new AI-based features emerge?
Can we personalise the customer experience?
Can we create new services from data we already hold?
Can AI partly replace services we sell today?

SoftwareAn AI assistant built into the product
IndustrialPredictive maintenance sold as a service
Professional servicesAI-supported client deliverables
RetailPersonalised recommendations
Financial servicesIntelligent customer advisory

Production

AI in how we create it

How does AI change the way we create our product or service?

Speed & Productivity

Speed
Output
Quality
Cost of production

Where does our value creation actually happen?
Which parts of it are still manual by default?
What would change if the same team produced twice the output?
Where does quality depend on a single experienced person?

SoftwareCoding agents, testing, code review, architecture support
EngineeringDesign assistance, simulation, documentation
ManufacturingQuality inspection, predictive maintenance, optimisation
Professional servicesResearch, document analysis, drafting, knowledge retrieval

AI does not only automate administration. It can change the economics of creating the product itself.

Organisation

AI in how we run the company

How does AI change the way we run the organisation?

Efficiency & Overhead

Efficiency
Automation
Administration
Decision support

Which internal friction is simply accepted today?
Which reports exist because someone once asked for them?
Where does management wait for information it could already have?
Which support functions could work with half the manual effort?

FinanceReporting, forecasting, invoices, anomaly detection
HRRecruiting, onboarding, employee support, training
SalesCRM hygiene, account research, lead qualification, proposals
MarketingContent production, market intelligence, analytics
Legal & complianceContract review, policy monitoring, document analysis
ManagementDashboards, decision support, scenario analysis

The second dimension

Replace · Augment · Create

Not a fourth bucket. A second layer that applies to all three. Every opportunity can be read in one of three modes.

Replace

Can AI take over an existing task or process entirely?

Report creation
Manual classification
Repetitive document processing

Augment

Can AI make a person substantially more capable?

Analyst
Salesperson
Developer
Engineer
Manager

Create

Can AI enable something that was not economically or technically sensible before?

A new AI-powered product
Personalised service at scale
Continuous customer advisory
New data products

Replace

Augment

Create

01Product

A self-service AI layer replacing a paid manual service step

Product features that make the customer better at their own job

A new AI-native offering built on data you already hold

02Production

Automated testing, classification and documentation

Engineers and specialists working with AI assistance by default

Delivery models that were previously uneconomic at your price point

03Organisation

Recurring reporting, invoice handling, first-line support

Management decision support and scenario analysis

Continuous internal knowledge access across every system

Illustrative examples. In a workshop, these fields are filled with your business.

Opportunity identification

The value is rarely where the demo is. It is in the repetitive, text-heavy, judgement-light work that surrounds your real business — and in the questions your managers cannot answer quickly today.

Organisational readiness

Readiness is not a maturity score. It is whether processes are described, responsibilities are clear and someone is allowed to decide. Companies that fail here fail regardless of the technology.

Adoption

Adoption follows leadership behaviour and a clear answer to one employee question: does this make my week better? Training alone has never produced it.

Agents and automation

Agents are useful where a process is stable, the systems are reachable and the failure mode is visible. Where those three do not hold, an agent is an expensive way to generate incidents.

Governance and data

One page that says what may be put into which tool, who approves exceptions and what gets logged. Written before rollout, not after the first incident.

ROI

Define the number before the project. Hours removed, cycle time, conversion, error rate. If nobody can name the metric, the project is a pilot — call it one and cap the budget accordingly.

Process redesign

Automating a broken process makes it worse at higher speed. Redesign first, then automate the version that people agree on.

Build vs buy

Buy the capability. Build the integration. Do not build what will be a standard feature in nine months — and be honest about which category you are in.

Employee enablement

Give people permission, examples and a safe boundary. Most organisations underestimate how much capability already exists and how little of it is being used.

What has to be considered?

The three buckets find opportunities. These dimensions decide whether they survive.

Once an opportunity has been identified in one of the three buckets, it is assessed across people, processes, data, systems and customers — plus risk and economics.

AI changes what an individual can produce in a day. That only becomes value when the organisation decides what to do with the time it frees.

Drafting, editing and translation as a default working habit
Onboarding new employees against internal knowledge
Specialist expertise available to non-specialists
Analysis work that previously required an external agency

Signature engagement

02Signature engagement

AI Opportunity Scan

When the workshop reveals a clear need for deeper implementation work.

Duration: 2–4 weeks
Input: Management interviews, process walkthroughs
Deliverable: Prioritised AI opportunity roadmap

A focused engagement that examines your company systematically across the three buckets: where can AI change customer value, differentiation or revenue? Where can it change how you create your product or service? Where can it reduce internal friction, administration and overhead?

Product — where AI can change customer value, differentiation or revenue
Production — where AI can change how the product or service is created
Organisation — where AI can reduce internal friction and overhead
Opportunities mapped across Replace · Augment · Create
Evaluated on business value, strategic importance, feasibility, time to value, data availability, risk and organisational impact

Every opportunity ends in one of four categories

NOW — implement
NEXT — test and prepare
WATCH — monitor
IGNORE — do not spend resources

Opportunity

Business value

Difficulty

Priority

Internal knowledge assistant
HighLowVery high
Proposal generation
HighLowVery high
Customer support AI
HighMediumHigh
Automated reporting
MediumLowHigh
CRM enrichment
MediumMediumMedium

Illustrative example of a scan output. Real matrices are built from your processes.

Explore the AI Opportunity Scan

Why this is ultimately a management question →

You probably don't need another AI presentation.

You need to know what AI means for your company and what you should actually do next.

Let's talk

30-minute introductory conversation. No sales presentation.