Decision-making and decision-making processes for AI in companies: From strategic impetus to practical implementation

More than just a tool: Why choosing AI will change your entire business

The hype surrounding artificial intelligence remains unbroken, and a gold rush mentality prevails in the boardrooms of German companies. Many see the introduction of AI as a quick, operational decision – just another software tool promising efficiency. But this assumption is a costly mistake and the main reason why a shocking 80 percent of all AI projects fail. The reality is: The decision to strategically integrate AI into a company is not a sprint, but a marathon that takes six to nine months before the first line of code is even written.

The reason for this complexity lies not in the technology, but in the process. Unlike conventional software, AI requires a fundamental reorganization of corporate strategy, governance structures, and risk assessment. Since the breakthrough of ChatGPT and the entry into force of the EU AI Act, non-committal experimentation is no longer an option. Every AI initiative today must be embedded in a rigorous legal, ethical, and financial framework.

This article is your guide through this demanding yet crucial process. It breaks down the complex path from initial strategic considerations to a decision ready for implementation into seven concrete, comprehensible phases. Using practical examples, cost analyses, and the most common pitfalls, you'll learn why the real work begins long before the technical implementation and how to set the course for a successful AI transformation – with strategic foresight rather than blind activism.

A strategic dilemma: Why AI decisions take longer than companies believe

The decision to introduce artificial intelligence into a company is often perceived as a quick operational choice. The reality is considerably more complex. An AI implementation decision-making process is not a single moment, but rather a nested sequence of strategic, operational, organizational, and technical assessments that takes between six and nine months before the first implementation phase even begins. While companies in other technology areas can work with established decision matrices, AI decision-making is fundamentally different: it requires not only the evaluation of technical parameters, but also the reinterpretation of governance structures, change management strategies, and risk assessments, which are often not yet institutionalized in this form within organizations.

The tragedy for many companies lies in their underestimation of the significance of this decision. AI is frequently equated with other software implementations in management discussions, even though its complexity is many times greater. This leads to underfunded projects, optimistic time estimates, and ultimately, the infamous failures documented in the literature: current research indicates that 80 percent of all AI projects fail. A large proportion of these failures are not technical, but rather procedural in nature. They arise because the decision-making process was not structured rigorously enough.

The historical development: From utopia to pragmatic governance

To understand today's decision-making process, it's necessary to examine the developments that led to it. The first wave of AI adoption in companies was characterized by euphoria and technological optimism. In the 2010s, AI was primarily explored by large tech companies and well-capitalized startups. Traditional companies were initially skeptical, and later hesitant. Decisions at the time were simple: external consultants were brought in, academic models were tested, and if something didn't work, the project was quietly abandoned.

This period of non-committal development ended abruptly with the publication of ChatGPT in November 2022. Suddenly, AI was no longer abstract and scientific, but tangible and ubiquitous. This led to a massive acceleration in expressions of interest from corporate boards. The second wave we are currently experiencing is characterized by regulatory pressure, competitive pressure, and the recognition that AI is strategically important. The EU AI Act, which came into force in August 2025, as well as similar regulatory frameworks in other countries, have fundamentally structured decision-making. Companies can no longer experiment without commitment; every AI initiative must be embedded in a legal and ethical framework.

The third dimension of this development is professionalization. Gartner reports that 75 percent of companies will be using AI by the end of 2025. This represents mass adoption. With this widespread adoption, of course, come standards, best practices, and governance frameworks that were previously unnecessary. Companies implementing AI today can draw on an established body of knowledge and experience, which makes decision-making more structured but also more complex. The decision-making process is not faster today, but more thorough and better documented. This is the central development that defines the modern AI decision-making process.

The core mechanics of the decision-making process

The decision-making process for AI in companies does not follow a universal scheme, but rather established patterns that emerge in more mature organizations. These processes can, however, be broken down into concrete phases, each with its own criteria, stakeholders, and criticality points.

The first phase is the strategic evaluation or assessment phase, which lasts between two and four weeks.

In this phase, the first question to be answered is: Where does our company stand with AI? This is done through a structured AI maturity analysis, in which executives from various departments – from IT and finance to business development – ​​are interviewed. The goal is to capture not only the technical readiness but also the organizational maturity. Companies that become anxious at this stage and want to quickly move on to the next phase are making a fundamental mistake. The assessment phase is the foundation upon which all subsequent decisions are based.

The second phase is strategy and goal development, which lasts four to eight weeks.

This is where the company defines what AI should be for its business. This is not primarily a technical question, but a business one. Examples of questions include: Should AI primarily enable efficiency gains or create new business models? Should it be integrated into existing processes or establish separate departments? Which industries or functional areas have the highest potential? This strategic clarification requires intensive discussions at the board level. Many companies underestimate the time this phase takes because they dismiss it as mere rhetoric. It isn't. Clarity about the company's vision regarding AI determines all subsequent decisions. Companies without a clear strategy end up with AI projects that lack tangible business value.

The third phase is use case identification and prioritization, which takes six to twelve weeks.

This is the operationalized version of the strategic phase. Here, concrete, business-result-oriented use cases are identified. The company gathers ideas from various departments: How could AI specifically help you? This collection is intentionally unstructured. A systematic prioritization follows, based on an evaluation matrix that considers factors such as business potential, technical feasibility, data maturity, and risk potential. The prioritization process is the most critical point in this phase, as it brings together optimistic business departments and realistic technical departments. Managing these tensions and arriving at a well-founded priority is a management skill, not a technical one. Companies that select their top ten use cases through simple voting will later waste time on unprofitable projects.

The fourth phase is the risk and compliance assessment, which lasts four to eight weeks.

This is a phase that was virtually ignored in the first wave of AI adoption (before 2023) but is now crucial. This phase evaluates: What regulatory requirements affect the planned AI applications? What data is required and what is its legal admissibility? What ethical questions arise? What liability and compliance risks emerge? Ideally, this phase is conducted by a team that includes lawyers, compliance specialists, data protection officers, and technical experts. This is not optional. Companies that skip this phase or conduct it superficially will create massive problems for themselves later.

The fifth phase is financial planning and business case development, which takes four to six weeks.

Here, concrete investment figures are compiled. The costs for AI implementation vary massively depending on the project scope. Self-service AI solutions can start at €4,000 to €25,000 per month. Custom developments range from €15,000 to €32,000 for a prototype and can reach €50,000 to €100,000 or more. Infrastructure costs, which can range from €500 to €15,000 per month depending on the cloud solution, are an additional factor. And then there are the hidden costs: employee training (€300 to €4,000 per person), change management, data preparation (which can account for 60 to 80 percent of the project budget), and continuous optimization. Enterprise AI projects in medium-sized to large companies can start with a budget of €250,000. Business case development is crucial here. Companies must not only demonstrate the investments but also the expected returns. A conservative ROI for AI implementation is 214 percent over five years; optimistic estimates can reach up to 761 percent. This range underscores the need for realistic assumptions.

The sixth phase is the organizational preparation and governance structure, which lasts four to eight weeks.

This is a phase that often runs parallel to others, but deserves its own distinct status. Here, the following questions are defined: Who makes decisions about AI projects? What governance structure is required? Is a Chief AI Officer necessary? How will AI be integrated into existing decision-making hierarchies? Large companies with more complex governance requirements establish an AI Governance Board comprised of representatives from business units, IT, compliance, HR, and finance. Smaller companies can handle this more informally, but should still establish clear lines of responsibility. This phase is critical because it gives the AI ​​initiative legitimacy and structure. Companies without clear governance later fail due to competing initiatives or a lack of accountability in decision-making.

The seventh phase is stakeholder mobilization and change management preparation, which lasts four to ten weeks.

This phase anticipates resistance and prepares the organization for it. The classic change management process for AI follows a proven structure: In the first two to three months, awareness is raised. Employees are informed that AI is coming, not as a threat to their jobs, but as a tool to empower them. In the following three to six months, a spirit of experimentation is fostered. Quick wins are demonstrated. Volunteer pilot groups are formed. The subsequent six to twelve months are dedicated to scaling. Best practices are documented, and training is institutionalized. Stakeholder engagement is crucial: 78 percent of executives see AI-supported decisions as a strategic advantage, but this is not automatic. This conviction must be won. Companies that skip this phase not only create implementation resistance but also long-term cultural problems.

Only after these seven phases, which together last between six and nine months, is the company in a position to launch concrete pilot projects. This is a critical point that many decision-makers misunderstand. They think that the decision to implement AI is the starting point for practical work. In fact, the decision itself is a six- to nine-month process, and only after that does the implementation begin.

Our EU and Germany expertise in business development, sales and marketing - Image: Xpert.Digital

Industry focus: B2B, digitalization (from AI to XR), mechanical engineering, logistics, renewable energies and industry

More about it here:

• Xpert Business Hub

A topic hub with insights and expertise:

• Knowledge platform on the global and regional economy, innovation and industry-specific trends
• Collection of analyses, impulses and background information from our focus areas
• A place for expertise and information on current developments in business and technology
• Topic hub for companies that want to learn about markets, digitalization and industry innovations

The status quo: Decision-making as corporate reality

The current state of AI decision-making presents a striking picture. On the one hand, there is the regulatory urgency. With the EU AI Act becoming a binding framework, European companies must embed their AI use in a documented governance system. This makes decision-making a compliance necessity, not just a strategic option. 77 percent of organizations are already actively implementing AI governance programs. This is not optional, but mainstream. This widespread adoption means that companies can draw on established patterns. The market for AI governance tools and consulting is growing by 36.7 percent annually and will reach a volume of $29.6 billion by 2033. This means that decision-making is more professionalized today than ever before.

On the other hand, decisions are more real and stakeholder-driven than before. 47 percent of organizations list AI governance as a strategic priority. This means that decisions are not made in IT departments, but at the board level. This increases the rigor of the process because boards typically have more formal decision-making processes than IT managers. While this is generally positive, it also leads to significant implementation delays.

Practical reality also reveals a fragmented landscape. Companies that successfully drive AI adoption follow a structured four-phase model: exploration (two to three months), standardization (two to four months), integration (six to twelve months), and finally, transformation. These phases are not optional or quick to complete, but fundamental milestones. Companies that skip or cram these phases systematically fail.

Another aspect of the status quo is the cost reality. Compliance expenditures for AI deployment projects average €344,000, while R&D costs are around €150,000. This represents a 229% cost increase for governance compared to development. This explains why decision-making takes so long: the decision itself has become expensive.

From practice: Two case studies of real decision-making

The first case study concerns a medium-sized Berlin-based e-commerce company with approximately 500 employees.

The company recognized that its logistics processes needed optimization. The traditional approach would have been to implement new software. Instead, an AI initiative was planned. The decision-making process took eight months. In the assessment phase, the existing logistics processes were mapped, data quality was evaluated, and the existing IT systems were assessed. It turned out that the data quality was significantly worse than expected. In the strategy phase, it was defined that AI should primarily be used to optimize delivery route planning. In the use case phase, seventeen use cases were identified and prioritized into four: route optimization, inventory forecasting, customer service automation, and fraud detection. In the risk assessment phase, it was determined that most use cases were unproblematic from a regulatory perspective, but the handling of customer data for fraud detection had to be documented in compliance with GDPR. In the finance phase, an initial budget of €150,000 for twelve months was defined. A dedicated AI task force was established. After eight months, the pilot project for route optimization was launched. After six months of pilot work (a total of 14 months after the initial decision), the results were measurable: an average reduction in delivery times of 18 percent and a reduction in logistics costs of 12 percent. These successes led to the project's expansion to other use cases.

The second case study concerns a multinational corporate holding company, RSBG SE, with over 80 subsidiaries.

The decision to implement AI company-wide took nine months. A critical difference compared to smaller organizations was the need to establish consistency within a highly decentralized structure. The assessment phase evaluated the AI ​​maturity of each subsidiary separately. It became clear that the maturity levels varied significantly. While some companies were already experimenting with AI, others were completely inexperienced. In the strategy phase, it was decided that AI should primarily be used to increase efficiency in administrative processes – an application with cross-functional relevance. Use cases were collected decentrally with central coordination. Eighty individual application ideas were submitted. These were categorized into quick wins (solvable in one to three months) and strategic projects (six to twelve months). In the risk phase, the central challenge was that compliance requirements differed across countries. A minimalist governance framework was developed, using EU requirements as a baseline. A central AI platform was selected. After nine months of decision-making, the scaling process began. Within three months, 60 percent of the companies were active on the platform. Over 80 use cases were identified and work began on their implementation. Within a year, AI saved over 400 hours per month. This is an example of successful scaled decision-making.

The problems and controversies: Where decisions fail

The central flaw in AI decision-making is unclear objectives. Many companies decide to implement AI without clearly defining what they want to achieve. They adopt AI because it's trendy, not because it solves business problems. This leads to projects without tangible benefits. Empirical evidence shows that 80 percent of all AI projects fail, and a large proportion of these failures are procedural, not technical. They stem from decisions made without a clear business objective.

A second key mistake is underestimating data quality and preparation. Many companies assume that AI systems can work with any data. The reality is far more critical. Typically, 60 to 80 percent of an AI project budget is spent on data preparation and cleansing. Companies that fail to anticipate this experience massive budget overruns and delays. Therefore, the decision to implement AI must always include a data quality audit.

A third key mistake is underestimating resistance to change and the need for cultural shifts. Many companies assume that if the technical solution is good, employees will automatically adopt it. This is psychologically naive. People fear that AI threatens their jobs, that their expertise will become obsolete, and that machine decisions will take away their control. A good change management program is not optional, but essential for success. Companies that underestimate this create technical solutions that fail in practice because employees don't use them.

A fourth mistake is inadequate project management and resource planning. AI projects are complex. They require technical expertise, domain knowledge, and project management simultaneously. Many companies underestimate the time and resources required. They assign AI projects as side jobs to employees who are already working at full capacity. This leads to delayed timelines and suboptimal results. Therefore, the decision to implement AI must always be accompanied by resource planning that anticipates realistic capacities.

A fifth critical error is the lack of success measurement and continuous optimization. Companies often fail to define measurably what success means. They launch AI projects without clear KPIs. This leads to a situation where, at the end of the project, it's unclear whether it was successful or not. Good AI decision-making defines measurable success indicators: time savings, cost reductions, quality improvements, and increased customer satisfaction. Without these definitions, the project becomes a political issue, not an empirical one.

Finally, there are the governance and compliance issues. The EU AI Act makes these issues non-optional. Companies that implement AI without evaluating their compliance requirements will create massive problems for themselves later. Particularly in regulated sectors (financial services, healthcare, insurance), the compliance phase is not optional. This also explains why the decision-making process takes longer than many companies expect: it must be defensible from a regulatory perspective.

The future of AI decision-making: trends and potential disruptions

The future of AI decision-making in companies will be shaped by several significant trends.

The first trend is the move from generative AI to agentic AI.

This means autonomous AI agents that not only provide recommendations but also make independent decisions and execute processes. This will fundamentally change decision-making. When AI systems not only analyze but also act, new governance requirements arise. Companies no longer have to decide what AI recommends, but how AI acts autonomously. This will make governance even more complex. Gartner predicts that by 2028, around 33 percent of all enterprise applications will integrate AI agents—a massive increase from less than 1 percent in 2024. This means that decision-making will not become faster in the coming years, but more complex.

A second trend is the democratization of AI.

No-code and low-code AI platforms enable not only technical experts but also business departments to develop AI solutions. This leads to decentralized AI adoption, which is harder to manage. This will change governance requirements. Instead of top-down decision-making, companies will have to deal with bottom-up AI initiatives. This could make decision-making faster, but also means a greater need for control.

A third trend is the integration of AI into existing business tools.

Microsoft 365 Copilot, Google Workspace AI, and similar integration options mean that AI is no longer a separate technology but an integral part of everyday tools. This simplifies adoption from a technical perspective but makes decision-making more complex because the lines between IT and business decisions become blurred.

A fourth trend is regulatory consolidation.

With the EU AI Act as an established standard and similar regulations in other jurisdictions, governance will become less fragmented. In the long term, this could standardize decision-making and thus speed it up. However, in the short term (the next two to three years), regulatory adaptation will increase complexity.

A fifth trend is the agency of AI decision-making itself.

It is expected that AI systems will not only support data analysis in the future, but also governance itself. Intelligent systems could simulate decision-making processes, run through scenarios, and assess risks before humans decide. This could improve the quality of decisions, but would also mean that decision-making itself is supported by AI – a reflexive paradox that raises its own questions.

What we can learn from this process

The decision-making process for AI in companies is not a single moment, but a structured process lasting between six and nine months, comprising seven distinct phases: strategic evaluation, strategy and goal development, use case identification and prioritization, risk and compliance assessment, financial planning, organizational preparation, and stakeholder mobilization. Only after these phases does the actual implementation begin. This timeframe deters many companies that dream of faster solutions, but it is necessary. Companies that accelerate or skip these phases systematically create operational problems for themselves.

The process is rigorous because the decision is critical. AI investments are strategically significant today. They can transform companies or lead them astray. Decision-making is therefore not a routine administrative task, but a core management competency. Companies that have successfully undergone AI transformations differ from those that fail not through technological superlatives, but through rigorous decision-making. They have defined clear goals. They have systematically evaluated risks. They have engaged stakeholders. They have defined success criteria. These management virtues are not new – they are simply explicitly required in the context of AI.

The future will show whether decision-making becomes faster or slower. Current dynamics suggest it will become more complex. With agentic AI, regulatory consolidation, and decentralized AI initiatives, governance requirements will increase, not decrease. Companies that anticipate this complexity will be better positioned than those that dream of fast, intuitive decisions. The key takeaway is: AI decision-making is not about speed, it's about accuracy. This is the central lesson for companies embarking on this journey.

Independent AI platforms as a strategic alternative for European companies - Image: Xpert.Digital

Ki-Gamechanger: The most flexible AI platform-tailor-made solutions that reduce costs, improve their decisions and increase efficiency

Independent AI platform: Integrates all relevant company data sources

• Fast AI integration: tailor-made AI solutions for companies in hours or days instead of months
• Flexible infrastructure: cloud-based or hosting in your own data center (Germany, Europe, free choice of location)

• Highest data security: Use in law firms is the safe evidence
• Use across a wide variety of company data sources
• Choice of your own or various AI models (DE, EU, USA, CN)

More about it here:

• Independent AI platforms vs. hyperscalers: Which solution is right for you?

Konrad Wolfenstein

I would be happy to serve as your personal advisor.

contact me under Wolfenstein ∂ Xpert.digital

call me under +49 89 674 804 (Munich)

LinkedIn
 

Benefit from Xpert.Digital's extensive, fivefold expertise in a comprehensive service package | R&D, XR, PR & Digital Visibility Optimization - Image: Xpert.Digital

Xpert.Digital has in-depth knowledge of various industries. This allows us to develop tailor-made strategies that are tailored precisely to the requirements and challenges of your specific market segment. By continually analyzing market trends and following industry developments, we can act with foresight and offer innovative solutions. Through the combination of experience and knowledge, we generate added value and give our customers a decisive competitive advantage.

More about it here:

• Use the 5x expertise of Xpert.Digital in one package - starting at just €500/month