AI is everywhere in wealth management: board agendas, strategy decks, vendor roadmaps. But while ambition is accelerating, the technology foundations beneath it often aren’t ready.
That was the central theme at A Meeting of Minds, a roundtable hosted by Owen James at The Berkeley Hotel, London, and facilitated by Milan Patel, Head of Sales Engineering at FA Solutions, alongside Sharmil Patwa of Opus Una.
The room brought together senior wealth management and private banking professionals to tackle one question: can wealth firms genuinely harness AI’s potential on legacy technology foundations, or does transformation require something more fundamental?
The insights below are anonymised and pooled. This isn’t one firm’s view: it’s the industry speaking with a collective voice. One line from the discussion set the tone for everything that followed:
"AI isn’t creating new problems. It’s exposing the ones we already had.", said by an event attendee – A Meeting of Minds, a roundtable hosted by Owen James.
It’s a dynamic that’s played out in other industries before wealth management: deploy AI into a fragmented, inconsistent environment and it doesn’t fix the fragmentation, it amplifies it.
Legacy foundations are the biggest barrier to AI adoption
Participants were consistent on this point. Most firms are still constrained by three things at once:
- Legacy technology estates – built for stability, not intelligence
- Fragmented, inconsistent data spread across multiple systems
- Operating models that were never designed for AI-driven workflows
Together, these constraints turn AI into an overlay rather than an integrated capability. Firms can generate impressive outputs: summaries, insights, recommendations, but struggle to embed any of it into core systems. The result is a growing collection of isolated AI tools that add complexity rather than remove it.
The takeaway participants kept returning to: AI cannot compensate for outdated infrastructure or poor-quality data. No model, however capable, fixes a bad data estate underneath it.
Firms are prioritising productivity and client experience, not efficiency
Three areas stood out as where AI is already delivering measurable value:
- Adviser productivity – reporting, analysis, workflow automation
- Client personalisation – contextual insights and tailored recommendations
- Investment decision support – scenario modelling, pattern recognition
What’s notable is what didn’t make the top of the list: operational efficiency. Despite arguably being the fastest route to quantifiable ROI, it ranked lower than the client-facing use cases. That’s a telling signal about how the industry currently frames AI, as a client-facing innovation story, rather than a structural, back-office transformation.
The risk in that framing is that the highest visibility use cases are often the ones that most depend on strong foundations to scale, and without them, even the most compelling use case stalls.
Integration is now the hardest part
A few years ago, the challenge was generating useful AI output at all. That’s no longer true. The challenge now is integrating that output into legacy estates without simply creating a second, parallel set of processes to manage alongside the old ones.
Participants pointed to four recurring friction points:
- Multiple data sources with inconsistent formats
- Systems that simply cannot consume or act on AI-generated insights
- Governance frameworks that were never designed with AI-driven decisions in mind
- Vendor ecosystems with uneven resilience and transparency
The unintended consequence is what several participants called AI silos: standalone tools sitting outside core workflows, technically functional but operationally isolating, adding friction rather than removing it.
Large enterprises face a different kind of complexity
Scale doesn’t make this easier, it changes the shape of the problem. Larger, more established firms are dealing with:
- Decades of accumulated legacy infrastructure
- Highly fragmented data estates across business units
- Complex governance and risk requirements
- Multiple business units with genuinely competing priorities
The firms navigating this most successfully are adopting what participants described as a dual model: grassroots experimentation to keep innovation alive at the team level, paired with central oversight to hold the line on governance, consistency and resilience. It’s a deliberate balance: loose enough to avoid stifling creativity, tight enough to prevent an uncontrolled sprawl of disconnected AI tools across the business.
Governance is becoming a critical enabler, not a blocker
AI adoption is forcing firms to strengthen oversight in five key areas:
- Data quality and lineage
- Model selection and explainability
- Cybersecurity and resilience
- Regulatory compliance
- Vendor due diligence
The framing that came out of the room was important: governance isn’t a brake on AI adoption; it’s the thing that makes adoption possible at all. Governance is a core enabler of trust, and without that trust, users quietly revert to manual processes and AI adoption stalls, regardless of how good the underlying tools are.
Modernisation is no longer optional
The roundtable’s clearest conclusion: incremental upgrades won’t close this gap. Patching or extending ageing platforms only delays a problem that’s compounding, not shrinking. Firms are starting to invest instead in modern technology environments built to support AI as a long-term capability, not a bolt-on.
The organisations most likely to succeed are the ones strengthening their technology and data foundations before trying to scale AI across the business, not the other way around.
Key takeaways
- AI cannot compensate for poor data quality or outdated infrastructure
- Adviser productivity and client experience remain the highest-value opportunities today
- Starting with clearly defined business problems delivers better outcomes than starting with the technology
- Integration with existing systems is now one of the biggest barriers to scale
- Governance, vendor resilience and cybersecurity are essential to any credible AI strategy
- Long-term competitiveness will depend on modern technology architecture, not AI tools bolted onto old ones
The bottom line: foundations first, ambition second
Wealth firms are right to pursue AI. The opportunity is real and the competitive stakes are rising. But the firms that succeed will be the ones that accept a simple truth:
AI is an accelerator. It amplifies whatever foundations already exist.
- Firms with fragmented data, legacy systems and inconsistent controls will scale inefficiency faster.
- Firms with modern architecture, clean data and strong governance will unlock the productivity gains AI promises.
The question for wealth firms is no longer whether they can innovate on legacy foundations. It’s whether they should, or whether it’s time to build the foundations AI actually needs.
Read the original article here.
