Artificial intelligence is no longer an experiment. In many organisations, it is becoming a part of everyday operations, boosting productivity, supporting decision-making and generating new business value. But how can we identify where AI holds the greatest implementation potential?
In this article, based on a talk by Piotr Modrzewski during a webinar co-hosted with Grant Thornton Poland, we explore not only how to pinpoint high-potential areas for AI, but also how to move effectively from idea to implementation, building lasting competitive advantage.
AI as a competitive advantage in business
In practice, many companies still ask themselves: is it worth investing in AI? The more relevant question, however, is how and when to do it. Artificial intelligence is no longer just a tool for optimisation, it is increasingly becoming a source of new business value. It enables the design of innovative operating models, supports data-driven decision-making, and allows automation of tasks that previously required human supervision.
The financial benefits are already evident: organisations that consistently invest in AI report significantly higher total shareholder returns than those that remain in the experimentation phase. At the same time, the competitive gap continues to widen the differences in technological maturity between leaders and laggards are growing year on year.
Where to implement AI to gain a competitive advantage
Not every organisation needs to automate everything. Success is not about implementing AI at all costs, but about accurately identifying the areas where the technology can deliver measurable results.
Based on research and market experience, several industries and business functions stand out in terms of effectively leveraging the potential of AI:
Compliance and Finance
AI delivers particular value here: automating reporting, detecting anomalies and ensuring regulatory compliance. Research shows that companies expect the highest returns on investment in these areas.
Technology and IT
AI supports development teams in writing and testing code and automates error analysis, enabling faster and more efficient creation of digital products.
Logistics and Supply Chains
From demand forecasting to route optimisation, AI drives savings, boosts efficiency and strengthens supply-chain resilience.
Marketing and E-commerce
Often the first area to adopt AI. It powers content personalisation, product-description generation, campaign automation and customer service, improving effectiveness and speed across the board.
Administrative Tasks and HR
AI accelerates CV screening, onboarding and document classification, relieving HR teams of routine work and allowing them to focus on strategic initiatives.
AI strategy in the organisation – key pillars
1. Vision and strategy
AI must form part of the business strategy, with a clear role, direction and C-suite sponsorship. Begin by identifying and prioritising low-risk, high-value opportunities (quick wins) to prove early benefits.
2. Organisational capability
Technology alone is not enough. Success depends on skills, organisational structure and culture. Companies need to build trust in AI, train staff to use it effectively and prepare for changes in roles and processes.
3. Experimentation and testing
AI deployment should be iterative. Starting with smaller steps such as Proof of Concept (PoC) projects or limited-scale pilots allows rapid validation in controlled settings, minimising risk while accelerating learning and improvement. Experimentation is not a threat but an essential stage on the path to scalable, organisation-wide solutions.
4. Scaling
The real challenge is moving from pilot to full roll-out. Successful solutions must be standardised, documented, monitored and embedded in business processes before they can deliver sustainable impact.
Organisational structures for AI
-
Central Model – Centre of Excellence (CoE) – Many firms establish dedicated AI centres to support other departments with project identification and delivery. A CoE promotes consistency, risk management and efficient use of scarce expertise.
-
Federated Model – Alternatively, AI capabilities can be distributed across business units. Data and AI specialists work directly with operational teams, ensuring solutions closely match real needs.
Our experience shows that multidisciplinary product teams, combining technical, analytical and business skills, significantly improve the pace and success rate of AI adoption. Such collaboration makes it easier to spot high-potential areas and translate operational requirements into concrete solutions.
Scaling AI solutions
One of the most common challenges is moving from pilot projects to full-scale implementation. Many AI initiatives stall at the Proof of Concept stage, failing to reach the production phase where they can deliver real business value. Effective scaling requires more than a well-developed model, it also demands changes to organisational processes, team workflows, and readiness to integrate new tools.
A key element is the industrialisation of AI, which involves standardisation, documentation, testing and ongoing monitoring of deployments. MLOps-class solutions play a critical role in maintaining the quality and continuity of models operating in production environments.
Integrating AI into business processes and training
-
Embedding into business operations – A successful AI implementation does not end with developing a model. It must be embedded into existing workflows to support day-to-day operational and decision-making processes effectively.
-
Evolving employee roles – The adoption of AI often leads to changes in roles, procedures and decision-making structures. This requires a redefinition of responsibilities and, in some cases, adjustments to motivational frameworks to reflect new ways of working.
- User Training – It is essential to prepare end users not only to operate AI tools, but also to understand how the models work and on what basis they make decisions. Such initiatives help build trust and support effective implementation.
- AI Academies – In response to the growing demand for AI-related skills, more and more organisations are setting up internal AI academies. Their aim is to retrain business-line employees and equip them to work effectively with data-driven and algorithm-based solutions.
AI-mature organisations are defined by a data-driven decision-making culture. Managers regularly rely on insights and recommendations provided by models, and trust in algorithms is embedded in everyday operations. Data and analytics form the foundation of decision-making processes at all levels of the organisation.
At the same time, a strong culture of experimentation is essential – a willingness to test new solutions and accept failure as a natural part of the learning process. Organisations that reach maturity in this area typically go through numerous iterations and are able to adapt their approach based on accumulated experience.
KPIs and success metrics for AI implementation
What exactly are we looking to improve by implementing AI? Most organisations focus on four core areas: cost, time, quality and productivity.
Cost efficiency typically means reducing operational costs (e.g. finance process costs as a percentage of revenue), achieving measurable savings both in percentage and absolute terms, and improving ROI metrics for AI projects.
Time efficiency refers to shortening process cycles (e.g. faster financial period closures, more efficient order processing) and accelerating decision-making, for example, through automated reporting delivered in hours rather than days.
AI also contributes to improvements in quality and accuracy, by reducing errors (e.g. fewer accounting corrections), increasing forecast precision and eliminating misclassifications.
Last but not least is productivity and throughput – more cases handled per employee, more time freed up for teams, and the ability to reallocate staff to higher-value tasks.
How mature is your organisation?
Before scaling AI solutions, it is worth evaluating your organisation’s current level of maturity. Not every business needs to operate like a market leader, the key is to assess your starting point realistically and choose next steps that align with your actual capabilities.
Data shows that only 1% of companies globally consider themselves fully mature in terms of AI with the technology deeply integrated into their processes and delivering measurable outcomes. A further 11% are executing strategic implementations across multiple areas, but without full integration. The majority of companies are still in the experimental stage or running isolated pilot programmes.
How to implement AI in your organisation – three approaches
1. The consulting/systemic approach
The systemic approach treats AI as part of a holistic transformation of the entire organisation. It requires full executive commitment, a clearly defined business objective, and a multi-year time horizon. It is based on AI maturity models, competency and technology gap analyses, and detailed implementation roadmaps.
This approach begins with a structured diagnosis: companies start by mapping their current state and defining what needs to happen to reach the desired level of maturity. Key components include AI readiness audits, market benchmarks, process and data architecture reviews, and organisation-wide capability building.
At Edisonda, we support this process with tools such as the AI Readiness Navigator – a comprehensive checklist that structures key questions, identifies gaps, and helps plan actions necessary to move from strategy to successful implementation.
2. The experimental approach
This research-driven model is based on iterative organisational learning, where artificial intelligence serves as a tool for exploration, hypothesis testing, and insight generation, rather than a finished solution. Implementation proceeds in phases, from pilots, through performance analysis, to decisions on scaling.
Organisations that adopt this approach systematically analyse empirical data and look for correlations between AI maturity and financial performance. Transformational leadership and a culture that encourages experimentation and learning from failure are particularly important.
In practice, tools such as the AI Canvas prove highly effective at this stage, helping teams organise ideas and define potential use cases for the technology.
3. The agile approach (design thinking)
In this model, AI is a component supporting solutions to real user problems, rather than a goal in itself. The approach is based on iterative testing, short learning cycles, and strong end-user involvement. The key is closeness to business and user needs, along with the ability to respond quickly to change.
The process typically begins with workshops and user research to identify high-potential areas for AI implementation. Agile teams work closely with stakeholders to ensure the technology delivers tangible value in real-world contexts.
The process often begins with workshops and user research aimed at identifying areas with high potential for AI application.
One of the tools that supports this stage is the AI Discovery Sprint – a comprehensive assessment designed to uncover potential and identify contexts in which artificial intelligence can enhance your product or solution.
Success factors in AI implementation
Even the most advanced technology will not deliver results without the right implementation environment. Successful use of AI depends on meeting several key factors.
First and foremost – strong executive support. Sponsors at the highest level play a vital role in securing resources, removing organisational barriers and fostering team engagement. Equally important is focusing on a clearly defined problem. AI should not be implemented for its own sake, but as a response to real business challenges.
Access to high-quality data is also crucial – it is the fuel for algorithms and determines the effectiveness of models. Organisations with large volumes of structured data have a natural advantage in this regard.
Another key factor is having the right competencies, whether in-house or sourced externally, that combine technological expertise with business context. And finally, change management is essential. Employees must understand and trust the new solutions, which requires consistent communication, education and support.
These are the elements that ultimately determine whether an AI project becomes a true driver of competitive advantage or remains just a pilot experiment.
