Innovation has been dominated by AI over the past few years. Firms have rushed to embrace the technology, experimenting with pilots and moving beyond into full deployment. While they needed to be quick, are firms embracing the technology correctly?
A report from EY in 2025 found that 95% of the 100 wealth and asset management firms it surveyed had scaled their adoption of generative AI across multiple use cases. Additionally, 78% were already exploring agentic AI tools to unlock more ‘deeper strategic advantages.’
In a similar finding, BlackRock recently stated that 68% of wealth management firms were using AI in some form. Of these, 34% were in a pilot, 28% had select use cases at scale, and 6% had multiple use cases at scale.
AI adoption is not just hollow talk at conferences, wealth firms are actively, and eagerly, embracing the technology into their workflows. FinTech Global recently spoke to Fincite CEO and co-founder Friedhelm A. Schmitt, Rob Paisley, AI Commercialisation & Revenue Innovation at SS&C Blue Prism and Fredrik Davéus, CEO and co-founder of Kidbrooke, to get their thoughts on whether AI is being embraced correctly.
The new must-have technology?
With the hype and speed of adoption, it is easy to see AI as the latest must-have technology, but is that the case?
Blue Prism’s Paisley believes there is a ‘must-have’ dynamic for the technology. He said, “Boards want an AI story, and not many CIOs want to be the one who didn’t act.”
That being said, Paisley noted that being seen as a must-have is not the same as something that is needed. Firms are going to derive more value from an AI solution if there is an actual problem that can be solved, rather than embracing the technology because everyone else is.
He said, “The firms getting this right are doing it strategically across the business, starting with specific, painful bottlenecks. That could be due to client data sitting in four different systems, advisers that are drowning in suitability paperwork, or a multitude of other reasons. That’s solving for optics, not for the client or the adviser.”
A similar opinion was shared by Kidbrooke’s Davéus. He believes that while there is a sense of indispensable surrounding the technology, there is a “fashion element” to it. He noted, “A good deal of the current spend is going into visible, marketable features rather than the parts of the value chain that actually determine outcomes.”
There are still problems within wealth management that need to be solved. Whether it is streamlining compliance, boosting efficiency or helping people make better financial decisions. AI can be a valuable solution to these problems, but only if implemented for the right reasons.
He said, “The discipline is to start from the customer problem and the economics, then work out whether AI is the right tool for it. It should never be a case of starting from the technology and then hunting for somewhere to apply it.”
fincite’s Schmitt was also sceptical of the notion AI being a must-have. But that doesn’t mean AI is not a great technology for wealth management when a firm actually needs it. He said, “In wealth management however it is the first technology that can make institutional memory operational. With AI wealth management firms will stop forgetting what they already know about their clients, their advisers and their own decisions. This is the huge value almost nobody talks about.”
Where an older tool is still perfectly fine
Fear of missing out is a powerful driver and can inspire teams to implement a tool that is too overqualified for a solution. Rather than looking at the minimum requirements to solve a bottleneck, firms might be tempted to leverage AI just because it is the exciting new solution. This is exactly what is happening, particularly with generative AI.
fincite’s Schmitt explained, “The industry is over-engineering AI wherever it uses intelligence to compensate for a lack of clarity. If a process is repetitive, stable and fully understood: Automate it. You don’t need AI for that. AI belongs where the answer cannot be predefined because context changes the decision.”
Overengineering solutions just to work with AI, such as generative AI, is something Kidbrooke’s Davéus sees as “one of the more expensive mistakes being made.”
He noted that generative AI is probabilistic, and finance is a domain where the answer is often needed twice. However, most of the use cases that is being pitched as finally fixed by generative AI, have been solved for many years by older technology.
He said, “Many of the use cases touted as “now solvable with AI” has been possible to solve for ages using classic technologies such as connecting systems using APIs, standardising and automating processes where additional flexibility is not required, etc. Also, working out whether someone can afford to retire, running a suitability check, projecting a portfolio across different market conditions: these are deterministic problems that call for a proper analytics engine, not a language model producing a plausible guess.”
He added, “Asking a large model to compute a figure that a well-specified engine produces exactly, every single time, is both costly and fragile.”
Finally, Blue Prism’s Paisley also offered some instances where generative AI is being overengineered within wealth management. The most common are in tasks that are rule-based with inputs that vary little. This includes extracting data from standardised KYC forms or checking transactions against fixed thresholds. He said, “deterministic automation will do it faster, cheaper, and with a more straightforward audit trail.”
Something that is far too common, according to Paisley, is firms using LLMs to summarise a document when really all that is needed is simple document extraction. While it might look great in a demo, it costs more to run and is harder to explain, which is a major problem for a regulated sector like wealth management.
He said, “Generative AI earns its place when there’s genuine judgement or unstructured language involved, drafting a first pass of client communication, interpreting a free-text query. Everywhere else, it’s usually just the expensive way to do something simple.”
Finding the right budget
Budgets are hotly contested for within businesses and there are various parts of the business where improvements and more budget allocation would be helpful. This leaves firms in a position where they need to decide whether their resources are better used on improving existing digital capabilities or investing into AI.
Kidbrooke’s Davéus is not sure these two areas will really be in genuine competition for the most part. He said, “Most firms do not have an AI gap, they have a data and analytics gap.”
Trying to put an AI layer on top of broken data will not bring that much value as it is only as powerful as the data beneath it. “If your customer data is fragmented and you cannot produce a consistent, accurate view of a person’s finances, a conversational layer on top will simply surface that inconsistency more fluently.”
His advice to firms is to first spend resources on improving foundations, such as a unified database, dependable calculation engine and clear customer journeys.
“Those investments pay off with or without AI, and they are the precondition for AI working at all. Get them right and adding an intelligent interface becomes a modest, high-return step rather than a leap of faith.”
This was a similar opinion to what Blue Prism’s Paisley had. A large percentage of wealth firms are running core processes on digital foundations that were built in a different era, he said. Regardless of how many AI tools that are layered on top, it will not improve a process that was never properly automated in the first place.
He added, “My advice to a firm with a constrained budget: fix the plumbing before you buy the smart tap. That doesn’t mean sitting AI out, it means being honest about sequencing. Shore up the workflows and data quality that any AI investment will depend on, then layer in AI where it adds judgement or speed that simpler automation can’t. Skip that step and you end up with AI projects that look great on a roadmap slide and stall the moment they hit real client data.”
If wealth management firms take the time to assess their infrastructure for real gaps where AI is the perfect solution for, they will then need to decide how to proceed. A firm can either spend resources and time to build a tool in-house, or they can look externally for a provider of a tool, but they will not own the technology.
fincite’s Schmitt does not recommend firms take the route of building the architecture in-house, with the advantage being from being quicker to market products. He said, “Under constrained budgets, wealth firms should not invest in building AI infrastructure. They should invest in becoming able to exploit it. The competitive advantage will not come from owning the model, but from adapting client journeys, adviser workflows and propositions faster than the market. The winning firms will buy the technology and own the speed of its application.”
AI use cases with lasting value
Whatever area wealth management firms choose to improve through with AI, there will be some that are successful and those that are not.
As to which use cases will prove to have lasting value, Davéus believes this will be mainly in interaction and explanation. Solutions that help customers understand their situation in plain language, answering follow-up questions that static questionnaires fail to handle and guiding people through decisions the moment they make it, will be powered up through AI.
He said, “That is where AI lifts the quality of advice and reaches people who never engaged with traditional channels at all.”
In terms of back-office operations, the technology will also be a big boon for operations, compliance and drafting, he added.
As for Paisley, the areas where AI will have lasting value will be “unglamorous.”
He explained, “Intelligent document processing for onboarding and KYC, decision support for suitability and risk checks, AI agents handling the back-office reconciliation and case management work that currently ties up skilled people. Those succeed because they’re measurable, you can point to hours saved, error rates down, time-to-serve improved, and because they’re built with governance and audit trails from day one, not bolted on after the fact.”
On the other side, there will also be areas where AI will not succeed and would have been a casualty of hype.
For Paisley, these will be around “fully autonomous “AI advisers”, or generative tools producing free-form advice with no human check in the loop. Regulators aren’t going to allow that, and clients with real money at stake don’t want it either. I think we’ll look back on this period and realise the value was never AI replacing judgement, it was AI clearing away everything that got in the way of judgement.”
As for Davéus, the unsuccessful implementation of AI will be in areas where it was aimed at replacing domain-specific financial reasoning.
He said, “What will look overdone in hindsight is the belief that a general-purpose model can replace domain-specific financial reasoning, along with the wave of undifferentiated assistants that all do roughly the same shallow thing. The firms that come out ahead will have paired conversational AI with rigorous analytics, clean APIs and correct data so the approachable interface is backed by numbers they can actually stand behind and prove.”
Read the daily FinTech news
Copyright © 2026 FinTech Global
Leave a Reply