Technological due diligence is increasingly crucial in M&A deals. How do you separate genuine AI capabilities from marketing spin?
The dream of the 10x AI-assisted software engineer dominates today’s M&A market and risks blinding private equity firms to both pitfalls and genuine opportunities. While investors chase dramatic claims of automated value creation, Wouter Denayer brings a more balanced perspective from years spent bridging Silicon Valley promise with boardroom reality.
Although deal origination still relies heavily on personal relationships, AI tools are reshaping how firms source new targets. "You cannot, as a team, detect every interesting company; it's impossible, especially if you're covering multiple countries", Denayer explains. Sourcing tools can help identify promising companies and generate targeted shortlists much faster. "You can cover much more ground, and the chance that you miss something becomes smaller."
AI can widen the funnel, but the differentiators remain distinctly human. Proprietary data sets built over years of relationship-building are what matters most now, and sharp investment theses guided by human intuition about where markets are moving next. "Finding the answer isn’t the hard part; figuring out the right question is.”
Denayer elaborates that the heart of dealmaking will always remain deeply personal. “A tool can’t build rapport and trust with founders or management teams, which is critical in today’s capital-rich environment.”
“A tool can’t build rapport and trust with founders or management teams."
Distinguishing marketing hype from technical reality
While the tools accelerate deal sourcing, they don't address the next, higher-stakes question: Is the target company's technology real, robust, and ready to scale? This is where the critical work of technical due diligence begins; separating genuine capability from clever marketing.
With 20 years at IBM and deep expertise in building software platforms, Denayer is well-positioned to separate genuine AI capabilities from marketing spin. "Four years ago, AI was maybe a paragraph in my technical due diligence reports; now it always gets its own chapter", he says. "Investors know they need an AI strategy, but distinguishing marketing hype from technical reality is almost impossible if you don't have deep technical and business expertise."
Now, an independent technology advisor for investors and their portfolio companies, his engagement often begins with technical due diligence. This involves validating what actually exists ("What do they really have?"), exposing gaps between pitch decks and reality ("The Information Memorandum is fundamentally a sales document"), and proposing clear solutions to help investors achieve their strategic goals.
"Few non-technical investors know where benchmarks lie, or how feasible any given claim might be compared with dozens of similar businesses already reviewed this year alone”, Denayer explains. “My role includes demystifying claims and explaining the fundamentals of technology for non-technical investment teams so they can better filter hype from substance and identify where real opportunities exist."
He draws parallels between greenwashing (in ESG) and today's rampant 'AI-washing': vague promises instead of qualified KPIs; hallucinations instead of measurable roadmaps; grand narratives instead of production use cases. "Don't believe what any vendor tells you at face value”, Denayer urges, but adds: “That doesn’t mean dismissing everything outright; just being more critical about whether results are genuinely being delivered."
‘AI-washing’ has become so pervasive that some firms have attempted mass layoffs, only to quietly rehire months later when the promised productivity gains failed to materialize. "If someone claims 10x productivity gains, it almost certainly isn't real…”
“In its current state, AI functions best as an amplifier for human judgement”, the independent advisor adds, pointing to language models saving days’ worth of manual review before meetings by summarizing large volumes of documents, for example; “It helps me prepare, but never replaces face-to-face conversations.”
"If someone claims 10x productivity gains, it almost certainly isn't real…”
The compound error problem – and why human oversight still matters
Wouter Denayer points out that we've been conditioned to believe fluency equals intelligence. “These systems sound smart because their language skills are super, but underlying errors compound quickly during multi-step reasoning tasks”, he says.
"Let's say an AI is wrong only one percent of the time. If you take its response and use that as the basis for a second prompt, any initial mistake gets carried forward and amplified”, he explains. "In a chain of prompts, these small errors multiply, in the same way that interest compounds."
With today's technology, this limitation undermines the entire premise of autonomous AI agents: the theoretical systems that would handle complex tasks independently. Denayer describes it as a domino effect: "A small error by one agent becomes the input for the next, and that error is then passed down the chain, amplifying with each step until you can end up with a major failure."
He adds that the "human in the loop" concept often fails in practice. "It becomes infeasible to manually validate every intermediate step of a complex, automated task. People tend to trust the final, well-written output, not seeing the compounded errors hidden within." For this reason, he advocates for embedding automated guardrails, such as validation against ground truth and business rules, to catch errors before they escalate.
For investors, it means you must go deeper during due diligence. Instead of just validating a company’s AI's claims, the ultimate responsibility sits with dealmakers to confirm the company has the technical maturity to build these essential guardrails. This means validating not just a feature's existence ("Is this feature live? Show me!"), but the system's architectural robustness, its security posture (too often neglected), and scalability plans. "There are endless rabbit holes in tech due diligence. You need discipline around what matters most for value creation after close”, Denayer says. “What's important to understand is that not everything has the same level of importance."
A new competitive moat
Denayer’s work with European mid-market companies reveals a sector in transition. "Most companies I see are in a growth phase, which goes along with growing pains", he says, adding that these companies face a dual challenge: scaling their core operations while integrating new AI capabilities.
The timing makes this particularly complex. "If you are five people, then you don't need many processes. But then you find yourself with 30 or 50 people – many are working remotely – and it's another ballgame”, Denayer’s explains. “Trying to add AI capabilities on top of these scaling challenges, with new people joining all the time, means you need much better processes.”
His role increasingly includes creating realistic AI roadmaps that focus on deliberate, incremental automation. "Let's understand what AI can do today and build from there", the tech expert explains. "It's about taking measured steps toward greater automation, deploying valuable features within the next 12 months, while being ready to adopt new capabilities as they mature. If you start projects aimed at impossible outcomes…you waste money and focus."
This has always been the adoption cycle of technology: “There's a shiny new tool, and then it's as if we are going to do everything with that new tool from now on.” Suddenly, businesses have to adopt all of these new tools (AI), and they don’t know how.
"One of my AI workshops focuses on education through historical context", Denayer explains. "I take people back to the origins of AI, walk them through how the technology actually works and end up showing all the powerful things we can actually do. When you understand the fundamentals, you can make informed decisions rather than getting caught up in either the hype or the fear. You need that foundation to distinguish between genuine opportunities and marketing narratives.”
"You need to understand the fundamentals to distinguish between genuine opportunities and marketing narratives.”
The right tool for the Job
Denayer advocates for a pragmatic "right tool for the job" approach. This means recognizing that while Large Language Models (LLMs) offer exciting new capabilities, they aren't always the optimal solution. In many business scenarios, the predictability of what might be called 'good old-fashioned AI' provides a key advantage. "Think classical machine learning models for forecasting or time-tested optimization algorithms for logistics", he says. "With these, there is no uncertainty, no hallucinations... they are testable, run on very small compute, and they don't cost a lot."
While new tools generate excitement, competitive advantage often lies in combining robust traditional approaches with judicious application of new ones.
AI will continue transforming deal processes, but success will increasingly depend on the ability to distinguish between genuine capability and marketing narrative. "Be critical", Denayer advises. "If it sounds too good to be true, then it is probably not true.
