Why we do not invest in AI promises, but in the layer beneath them, and why our bondholders benefit.

Everyone wants to do something with AI. Almost no one has their data in order.

Walk into ten Dutch boardrooms today and nine of them will have AI on the agenda. Copilots, agents, generative models in customer service. The ambition is there, and often the budget too.

What is usually missing is the foundation capable of supporting that ambition.

The data is scattered across systems that were never designed to work together. Definitions vary by department, “customer” means something different in sales than in finance. Historical data is incomplete or cannot be reproduced. Ownership has not been assigned. Data quality is measured when a problem occurs, not as a matter of routine. And governance exists on paper, not in practice.

Then the model arrives. And the model does what it always does: it amplifies whatever you put into it.

The oldest rule in model building

I have been working with data for more than twenty years. Every model I have seen, statistical, machine-learning or generative — is governed by the same rule:

Rubbish in, rubbish out. Diamonds in, diamonds out.

A mediocre model trained on excellent data outperforms an excellent model trained on mediocre data. Almost every time. That is not an opinion; it is the experience of anyone who has ever had to explain a production model to a client who did not trust the answer.

The implications of this are widely underestimated. Models have become a commodity: they are available to everyone, at falling costs and with narrowing differences between them. What is not interchangeable is the data layer underneath, the infrastructure, quality, definitions, history and ownership.

That is where scarcity lies. That is where the defensible position is built.

Why data management is a strategic asset, not a cost centre

In many organisations, data management sits under IT operations. It is treated as a cost line, not as value creation. That historical mistake is now becoming painfully visible.

Three reasons we treat it as an asset:

1. It cannot be copied. A competitor can buy the same model tomorrow. What they cannot buy tomorrow is fifteen years of clean, connected and documented data history.

2. It sets the ceiling for every AI investment. Skip the data layer and you pay twice: first for the project that fails, then for the foundation you still have to build.

3. It has become a governance imperative. Traceability, explainability and demonstrable quality, once considered good practice, are now prerequisites for operating in today’s regulatory environment.

Data management is not merely a prerequisite for the AI strategy. It is the AI strategy — in the only area you actually control.

What this means for where we invest

Liplyn Group does not invest in AI promises or in start-ups that have yet to prove there is a market. We invest in established Dutch data companies: businesses generating revenue today, with customers who have stayed for years, and with services or products in data management, data infrastructure, data quality and decision intelligence.

The profile we look for is consistent:

•     Proven track record. Several of our operating companies have been around for fifteen years or more. That is not nostalgia, it is accumulated data, customer relationships and expertise that cannot be recreated in two years.

•     Enterprise customers with long-term contracts. Sales cycles of nine months or more are a barrier to newcomers. For the incumbent, they form a moat.

•     Untapped demand. In virtually every company we acquire, existing customers ask for more than is currently being delivered. That is the least expensive growth available.

•     Succession opportunity. Many of these businesses were built by founders who are ready for the next chapter. We provide continuity for the company and its team, along with a structured transition.

Within the Group, we are building a portfolio whose parts reinforce one another: an information platform with broad reach and years of Dutch open and corporate data; a data management company that builds infrastructure and data quality practices for enterprise clients; segmented household and address data used by public authorities and market participants; and a property and valuation data layer currently under development.

The strategy has two phases. Phase one is the connected data layer: turning separate data sources into a single, coherent foundation for data and decision intelligence. Phase two translates that foundation into industry-specific, data- and AI-driven workflow solutions for customers. Phase one without phase two is infrastructure without margin. Phase two without phase one is exactly the mistake the rest of the market is making today.

The bond: investing alongside us in the foundation

We finance this buy-and-build strategy with a combination of equity and bond capital. For investors seeking exposure to the data economy without the risk-return profile of venture capital, we offer bond issue series 2026-III.

Key terms

Interest	9.0% per annum from € 1,000 · 9.5% from € 50,000 · 10.0% from € 100,000
Payment	Quarterly
Term	5 years
Repayment	Principal repaid in full at the end of the term
Security	First-ranking right of pledge over the assets
Flexibility	Convertible into shares and transferable
Bondholder representation	Through an independent foundation that represents the interests of bondholders

The interest is not funded by projections, but by the cash flow of companies already invoicing customers today. That is the fundamental difference from investing in the AI hype: our underlying businesses earn money from the work that has to be done before AI can create value anywhere.

More than forty investors are now investing alongside us.

In closing

The market currently pays a great deal for the model and very little for the data. That is exactly the kind of asymmetry from which returns are generated. We are buying the undervalued part: the expertise, infrastructure and datasets on which every AI application ultimately depends.

The post Rubbish in, rubbish out. Diamonds in, diamonds out. appeared first on .