Artificial Intelligence (AI) is increasingly playing a key role in data-driven business models. For investors and M&A professionals, this means: traditional tools of software due diligence are no longer sufficient to assess the risks and potential of modern AI systems.

In this blog of the blog series on "Software Due Diligence: The Key To Successful M&A Deals", we focus on the key aspects of AI due diligence regarding self-developed models (such as own Machine Learning models).

The emerging questions related to the use of large language models will be addressed in the next blogs.

Why Traditional Software DDs Fall Short for AI

While traditional software is based on fixed rules, AI software learns from training data and develops its decision logic independently. This means that the predictive performance of self-developed models depends not only on the code, but heavily on the quality of the data, the training process, and data preprocessing. These components are often not adequately considered in traditional software due diligence.

New Questions in AI Software Due Diligence

A thorough AI due diligence must address new types of questions:

• Has an AI model really been developed, trained, and deployed – or is it just a rule-based system?
• What was the quality and diversity of the training data?
• Are there monitoring, re-training, and fallback strategies in place?
• Is AI expertise available in-house or outsourced?
• How is bias in data identified and mitigated?

These questions are essential to evaluate whether the AI advantages claimed in the business plan are realistically achievable – or whether there's a risk of technologically “flying blind.”

Identifying Technological Risks

Poorly maintained or inadequately trained AI can quickly become a liability – whether through faulty predictions, lack of scalability, or overdependence on individual experts. Typical risk areas include:

• insufficient predictive performance,
• lack of bias monitoring,
• non-reproducible training processes,
• or lack of data access after a carve-out.

The latter in particular is often overlooked in traditional due diligence processes, but is crucial for the operational continuation of an AI-driven business model.

Not Everything Labeled "AI" Is Truly AI

Surprisingly, studies show that around 40% of analyzed “AI startups” do not actually use real AI. Therefore, it's essential to critically examine model architecture and training data. Only then can one assess whether the system truly meets the requirements of the use case – or whether traditional methods might be more suitable.

Organizational Maturity as a Success Factor

Technology alone isn’t enough. Successful use of AI strongly depends on whether:

• dedicated AI roles are established within the company,
• processes such as monitoring, retraining, and data management are in place,
• management and business units understand and support the significance of AI.

Companies with clearly defined AI responsibilities, budgeting, and roadmaps (so-called "AI High Performers") achieve significantly higher efficiency and performance improvements, according to studies.

Conclusion: New Assessment Standards Are Required

Artificial intelligence can create true competitive advantages in M&A projects – or pose significant risks. Traditional software due diligence is not enough to evaluate these risks. A specialized AI due diligence is required – with a focus on data quality, predictive performance, technological maturity, and organizational integration.

AI is like a sharp sword: those who harness its potential can lead – those who underestimate its risks may quickly get cut.

Act Now: Integrate AI Expertise Into Your Transaction Evaluation

Whether you're acquiring a tech company or investing in a data-driven business model – if AI plays a role, you need new criteria, new methods, and partners with hands-on experience.

👉 Cape of Good Code combines deep technology analysis with AI-specific expertise – delivering clear insights on the viability, scalability, and sustainability of your target technology in no time.

📞 Schedule a non-binding initial consultation or request our AI Software Due Diligence.

Links

[0] Photo by Markus Winkler