In an era where AI-driven wealth advice is more common and widespread than ever, being able to sort the beneficial from the not-so-helpful can sometimes be a challenge. Abbove, a Belgian WealthTech, sees itself as being able to provide a significant answer to this challenge.
According to Guillaume Desclée, CEO and co-founder of Abbove, the company is a ‘two-sided coin’, with this duality being a key selling point that sets it apart from competitors.
“The first side is a collaborative wealth planning platform,” said Desclée. “It enables private banks, family offices and advisory firms to scale personalised, family-centred wealth planning – anchored in each client’s life goals – across their entire client book. Adviser and client build together a structured view of the wealth: who holds what, through which entities, with which objectives, for which generations. It is an engaging experience because it speaks to what truly matters to families.”
Currently, 1,200 advisers from leading private banks, family offices and accountancy businesses across Europe use the Abbove platform to support over 40,000 families and 200,000 family members in creating and planning their wealth strategies. The company says its client report an adviser NPS of 65, and an 18% increase in end-client satisfaction.
The second side of the coin, Desclée outlines, is less visible but equally strategic. “Every interaction on the platform produces and enriches a structured, auditable and continuously updated wealth data model. Family structures, matrimonial regimes, ownership chains, transmission objectives, scenario simulations – everything is organised into a coherent system of context,” he said. This, the Abbove CEO states, is precisely the data layer that FIs are seeking today in order to deploy AI reliably within their advisory services.
Desclée remarked, “What makes this combination unique is that both sides reinforce each other: the platform generates the data, and the quality of the data makes the platform irreplaceable. In a world where agentic AI will progressively act upon portfolios and wealth strategies, the real question is not which model to use – it is the context upon which that model will reason. Abbove is built to be that answer.”
Structured context
For Desclée, platforms that sit above existing systems to organise and govern data before AI acts upon it are decisive – and is area where the industry vastly underestimates the scale of the challenge.
He said, “Most financial institutions hold abundant data: CRM, PMS, core banking, compliance tools. But this data is fragmented, duplicated, and above all lacks the semantic structure that AI requires to reason. Knowing that a client holds £2 million in a life assurance policy is not enough. You need to know that this policy is held through a Luxembourg holding company, that it was taken out to protect a dependent spouse, and that the named beneficiary is in open dispute with the other heirs.”
It is this level of context that Desclée believes transforms a generic AI into a useful advisory assistant.
“At Abbove, we have spent years building this wealth context model — not as a feature, but as the very foundation of the platform,” he said. “Every piece of data captured is structured around four dimensions that no traditional banking system models: family structure, asset ownership arrangements, applicable civil and succession law frameworks, and clients’ life goals. This is what we call the System of Context. Platforms that play this role are not merely important – they are a prerequisite.”
Connecting fragmented systems
A vast number of institutions still operate on fragmented legacy architectures. A key question facing many in the industry is what are the realistic paths to connecting these systems to the data environments that agentic AI requires? Here, Desclée believes the most realistic path comes not from replacing existing systems but building a sematic layer above them.
He said, “Core banking migration projects take an average of seven to ten years and cost hundreds of millions. No institution can make its AI strategy contingent on the completion of those projects.”
Desclée suggests here a three-stage approach. Firstly, connect existing systems through standardised APIs to aggregate available data without displacing it. Following this, enrich that data with the information that only a clientadviser engagement can capture, such as goals, intentions and family structures, which no transactional system stores. On the last step, firms must structure everything around a wealth data model that serves as a shared reference for all systems, including future AI models.
“This is a logic of progressive enrichment, not brutal replacement. And it delivers tangible results within the first few months, which is essential for maintaining team buy-in and board confidence,” said Desclée.
The human-AI boundary
The rise of agentic AI in wealth management raises a fundamental question: where should autonomy end and human judgement begin? In a profession built on trust and fiduciary duty, that boundary cannot be purely technical – it must sit where context and responsibility are most acute.
For Guillaume Desclée, the principle is clear. “The boundary must be drawn where judgement is required,” he says – particularly where outcomes depend on context that cannot be fully encoded, or where the consequences of error are irreversible. Succession planning, tax structuring and suitability assessments all fall into this category.
But Desclée cautions against a simple automation divide. “A sound wealth planning decision is not always the fiscally optimal one,” he explains. “It is the one that combines tax optimisation with family optimisation — and the latter is deeply subjective.” A technically perfect solution can still prove humanly damaging — an asset split that reignites conflict, or an ownership structure perceived as unjust.
As he puts it, “What AI cannot encode is a human reading of a family situation — the adviser’s sense of what is left unsaid, their understanding of what a family can genuinely live with together.” It is precisely here that human judgement must prevail and AI must step back.
The same logic applies to accountability. “The framework must evolve towards a logic of auditability rather than mere imputability,” Desclée argues. The real test is not simply who is responsible if AI is wrong, but whether every recommendation can be traced, explained and challenged. “Without full traceability of AI reasoning,” he concludes, “no accountability framework will be credible to regulators or to clients.”
How firms gain competitive advantage
Looking ahead to the near future, will competitive advantage come from better AI models or will it be achieved from the institutions that build the data architectures capable of deploying them safely?
For Desclée, the question is already settled, even if, in his words, many institutions have yet to internalise what it means for their strategy.
He said, “Foundation models are converging rapidly: GPT, Gemini, Claude are reaching comparable levels of performance across most generic tasks. Within two to three years, accessing a powerful model will be as simple and as undifferentiating as accessing an internet connection.”
What will remain scarce and difficult in the opinion of Desclée is the context upon which those models will operate.
He concluded, “Competitive advantage will not come from the model. It will come from the depth and reliability of the context upon which the model operates. The institutions that have understood this are investing today in their wealth data architectures. The others will catch up – but at a considerably higher cost.”
Copyright © 2026 FinTech Global
