Robert Kelly, Co-Founder, Graphene
The wealth management industry is having the wrong conversation about AI. Firms are debating which tools to adopt. The more consequential question is what those tools are sitting on.
I am not arguing against AI in the advice process. The efficiency gains are real. Suitability documentation that once took hours is produced in minutes. Annual review cycles that ran for weeks now close in days. Compliance monitoring that previously required a paraplanner in the room now runs without one. I expect these capabilities to be standard practice across the industry within eighteen months.
The technology works. That is not what concerns me.
What concerns me is a structural fact that tends to get lost in the pace of adoption. Every AI product in this category describes itself in similar language. Integrates with your CRM. Connects into your back office. Works across any level of data maturity. What that language consistently describes is an intelligence layer built on a foundation held by somebody else. The client record, the asset register, the custody permissions, the regulatory authorisations that make any of it lawful. None of that belongs to the AI vendor. It belongs to the platform beneath it.
That is the fact I think deserves more attention than it is currently receiving.
Why Wealth Management Is Different
In most software categories, sitting above the infrastructure is a perfectly sensible position. A reporting tool does not need to own the data it presents. A CRM does not need to own the database it queries.
Wealth management is different because the foundation is regulatory, not just technical.
Custody permissions cannot be rebuilt over a weekend or engineered around when a platform relationship deteriorates. A firm that owns its permissions owns its operational continuity. A firm that does not carries a dependency it cannot remove by switching a tool. Layer an AI system on top of that dependency and you have compounded it. The component that looks most modern turns out to have the weakest structural claim on the business.
The economics follow directly. An AI layer is only as valuable as its access to the underlying data, and that access is granted, priced and revocable by whoever owns the platform. A platform does not need to act in bad faith for this to matter. It only needs to act in its own commercial interest. Change the API terms and the AI layer becomes more expensive. Restrict the data feed and it performs worse. Bundle a competing workflow product at no extra cost and the value of the layer above it diminishes considerably.
Regulatory accountability reinforces the point. Serious AI vendors acknowledge it themselves. Generated documentation stays subject to adviser review, and the firm retains oversight. Under Consumer Duty, liability for a suitability outcome sits with the regulated firm. The accountability cannot travel upward to the AI layer because the AI layer holds no permissions and carries no regulatory standing. The weight sits where the platform sits.
What Consolidators Will Discover
The consolidators will feel this most acutely, and at the worst possible moment.
Acquire several businesses across different platforms and the AI tool standardised across the network integrates cleanly with some of them, partially with others, and the roadmap for full support on the remainder belongs to a company with no particular obligation to prioritise your timeline. The efficiency case that justified the investment dissolves into integration work. The problem is not the AI. The problem is that the platform was never unified, and no amount of intelligence above it can substitute for that.
Data sovereignty is not a technology aspiration for consolidators. It is the prerequisite for scalable, consistent operations and the enterprise value that follows from them.
“The AI era in wealth management is not a race to install the newest layer. It is a decision about what that layer is going to sit on. The firms that make that decision deliberately and early, will find it was the most important one they took.”
Robert Kelly, Co-Founder, Graphene
What Ownership Actually Changes
At Graphene, custody, permissions, the client record and the asset register sit inside the platform rather than being accessed through it. That single structural fact changes the value of everything built on top.
It means there is no third party who can reprice access or restrict a data feed that a firm’s operating model depends on. The commercial relationship sits directly between Graphene and the firm.
It also means we do not compete with the firms we serve. A platform that runs a proprietary advice business, or that has an asset management operation to support, has an interest in what sits on its infrastructure. That interest does not need to be malign to create misalignment. Graphene does not manufacture product and does not distribute advice. The incentive to move up the stack and compete with our clients simply does not exist, and firms building AI capability on our platform can do so without asking what their infrastructure partner’s next move might be.
The Decision That Will Define the Next Decade
I want to be honest about the limits of this argument. Owning the platform is a durable advantage only for as long as what is built on it continues to deserve that confidence. The obligation runs both ways.
But the firms that will compound their advantages in the AI era are not those that adopt the most tools. They are those that understand which parts of the stack can be sourced externally and which parts must be owned. They will treat the client record, custody permissions and operational data as the assets they are, and choose infrastructure partners whose interests are aligned with theirs rather than adjacent to them.
The AI era in wealth management is not a race to install the newest layer. It is a decision about what that layer is going to sit on. The firms that make that decision deliberately and early, will find it was the most important one they took.
Are you evaluating AI adoption? Get in touch
If you lead or advise a wealth management firm evaluating AI adoption and want to understand how platform ownership affects the value of every layer above it, we would welcome a conversation. Get in touch via the button below.


