Blog post by Christian Schipp, Partner
AI disillusionment: Why 95% of projects fail—and how we can finally create real value
Many companies are now facing reality: despite enormous hype, current studies show that around 95% of all company-wide AI projects fail. The problem is not the technology—it is more powerful today than ever before. The decisive factors are other things: a lack of strategy, insufficient integration into processes, and inadequate implementation.
What do studies show?
→ According to a recent study by the Massachusetts Institute of Technology (MIT), around 95% of all company-wide AI projects fail. Only around 5% of the initiatives examined lead to real added value – in other words, measurable business results. The reasons given are: lack of context, poor coordination of AI applications with day-to-day business, and insufficient integration. (IT-Matchmaker News)
→ A study for the German market makes it clear that failure is often not due to the technology itself, but to poor integration into processes and organizations. (IT-Matchmaker News)
→ In reality, this means that many pilot projects remain on paper or in the test phase – without ever being put to productive use.
What do consulting firms show?
→ According to a 2023 report by McKinsey & Company (“The economic potential of generative AI”), generative AI—combined with other technologies—could theoretically enable productivity gains of 0.2 to 3.3 percentage points per year. (McKinsey & Company)
→ At the same time, recent McKinsey surveys show that although many companies are investing in AI, the measurable economic benefits are often lacking: over 80% of companies report no significant gains from AI use. (Observer)
→ However, there are also examples where AI is already delivering benefits in certain functions, such as marketing or supply chain. (McKinsey & Company)
Conclusion: Although there is enormous theoretical potential, the vast majority of AI projects have not yet realized these benefits, whether for strategic, organizational, or technical reasons.
Why does this happen? → Typical sources of error in companies
Based on the experiences of many companies—and also from the everyday work of consulting firms—several systematic causes can be identified. The most common are:
→ No overarching strategic vision for AI: Many companies see AI as just another function or software component (“we need ChatGPT or Copilot”), rather than as a lever for profound change: Business models, processes, and culture remain unchanged.
→ Inadequate foundations at the data, process, and application levels: Data quality, process definitions, and system landscapes are often insufficient or fragmented. Without clean data and stable processes, you get “garbage in, garbage out”; AI delivers results, but they are unreliable or hardly usable.
→ Overambitious, unspecific approach instead of focused use cases: Some companies try to cover various areas with general AI initiatives (“AI everywhere”) – without clear prioritization. Then little happens except that resources are wasted.
→ Lack of technological and organizational expertise: Some service providers sell AI workshops and workshops with great hype, but without having in-depth experience in the architecture of business applications (CRM/ERP) or integration into existing IT landscapes. As a result, initiatives often remain at the level of a prototype or a playground.
→ Lack of willingness to change or cultural barriers: AI changes ways of working—those who do not adapt processes, roles, and responsibilities will hardly see any lasting effects. Resistance, a lack of clear governance, or inadequate change management prevent success.
My assessment of the current situation
“Personally, I think it's understandable that hype and reality still differ greatly today – but with the right approach, this gap can be closed in a targeted manner.” – Christian Schipp, CBDO & Partner
Technologically, AI is widely available today – generative models, agents, and automation tools have reached an impressive level of maturity. But the art lies not in the technology, but in its integration: those who see AI only as a “tool” and not as a strategic lever will be disappointed.
Many companies are jumping on the bandwagon because the pressure is mounting (“everyone is talking about AI”) – without seriously questioning their organization, database, and processes. This leads to many half-baked, ineffective, or meaningless projects.
At the same time, I believe that the “5% success stories” just described are precisely those companies that take a bold but structured approach: with a clear vision, a focus on concrete processes, clean data and system architecture, and a willingness to tackle change holistically.
Therefore, AI hype and AI success are not the same thing. Success means discipline, patience, and long-term commitment.
How we position ourselves at Ambit Group – our approach to effective AI in business
Here we show how we at Ambit Group make a difference – why we are convinced that we can enable real, sustainable value creation:
→ Focus on business processes & business applications (CRM, ERP, custom software) We don't see AI as an end in itself, but as a lever within existing ERP/CRM processes – where business results are generated. Our experience with business applications helps us identify meaningful AI use cases.
→ Holistic integrations — even under the hood (e.g., business applications, Azure, data architecture, end-to-end processes): We offer not only consulting, but also technical implementation and ongoing support — from prototype (PoC) to productive use. This allows us to avoid “lighthouse projects” that never make it past the pilot phase (no blah blah blah).
→ Methodical approach with clear use cases, PoC phase, and step-by-step scaling: Instead of broad-based experiments, we start with concrete, tangible use cases (e.g., in sales, customer service, finance), test them in PoCs, measure the effect, and scale systematically. Here is the link to the PoC workshop PDF.
→ Many years of experience in the business application environment: Thanks to almost 30 years of practical experience in the field of business applications, we know the pitfalls and what matters—processes, data quality, change management, integration.
→ Commitment to sustainability and results: We don't focus on short-term effects, but on sustainable, measurable productivity gains — realistic, transparent, and based on target agreements with the customer.
With this approach, we are convinced that — unlike many “AI playgrounds” — we can realistically achieve productivity increases of well over 10–20% if the framework conditions and goals are clearly defined. Here is the link to the AI workshops.
Why now is the right time — and how to get started
→ The pressure is mounting: the market, customers, and employees increasingly expect AI expertise. Companies that fail to embrace change risk falling behind their competitors.
→ But unlike previous IT waves, AI doesn't just impose technology — it requires organizational, procedural, and cultural adaptation. Those who get started now with a sense of proportion, methodology, and a clear vision will lay the foundation for sustainable transformation.
→ Our structured approach via workshops → concrete use cases → PoC → productive use → support is ideal for medium to large companies that want to use AI sensibly — without getting carried away.
The Microsoft platform provides the foundation for effective AI projects
Today, Microsoft's technological base provides a robust and scalable platform on which companies can reliably and efficiently build AI solutions. This is crucial—because technology alone does not make AI successful, but without a stable foundation, it hardly works at all. Here are the main reasons:
Enterprise-class platform for AI in businesses
→ With Azure AI Foundry, Microsoft offers a comprehensive environment for the entire AI lifecycle management: from model selection to training, deployment, monitoring, and governance. This enables companies to develop, operate, and scale individual AI solutions—with reliable control over data, security, and compliance.
→ Through services such as Microsoft 365 Copilot or integrated AI functionalities in Dynamics 365 and Power Platform, Microsoft already provides highly productive solutions “out-of-the-box” — for example, automatic text or conversation summaries, intelligent suggestions in emails, documents, or CRM data.
→ For individual requirements, tools such as Copilot Studio or AI Builder can be used to build customized agents or workflow automations — such as chatbots, document processing, or process automations — deeply integrated into existing business processes and applications.
Flexibility and integration into existing business applications
→ The combination of Azure Cloud, Dynamics 365, Power Platform, and Microsoft 365 allows AI to be seen not as a stand-alone solution, but as an integral part of corporate IT: data, processes, and user interfaces remain connected—this avoids media breaks and fragmented systems. At Ambit, this is exactly where we come in. Microsoft Copilot only delivers the desired effect when it is embedded in the entire application landscape.
→ The modular architecture of Azure and AI services allows you to start small (e.g., with a pilot in one area) and scale gradually if successful — without turning the entire company upside down. This keeps the effort manageable and the benefits measurable.
Security, compliance, and governance—important for businesses
→ Microsoft's AI platform is not only technologically powerful, but also enterprise-ready: security, data protection, and governance are taken into account from the outset—for example, through access controls, compliance certifications, service management, and clear role models.
→ This is a decisive advantage over many pure AI startups or isolated solutions, especially for business applications that process sensitive customer data, financial data, or confidential documents.
A wide range of tools for many use cases – from standard to highly customized
→ For simple, quick improvements, there are ready-made “Copilot” functions in Dynamics 365 & Power Platform – ideal for sales, customer service, marketing, office production, etc.
→ For more complex requirements—e.g., special workflows, document automation, cross-process applications, or special solutions for industry, non-profits, retail, associations, etc.—Azure + Copilot Studio + Power Platform enable implementation with individuality and a high degree of integration.
Are you interested in this topic? I would be pleased to answer any queries you have.
Christian Schipp, Chief Business Development Officer
christian.schipp@!ambit-group.com
+41 79 954 17 09
