Wealth Dynamix

Named model features sit on top of a client lifecycle management platform that would survive their removal intact: a persona builder, a prospect to adviser matching tool, predictive analytics and sentiment analysis, running over a customer relationship, onboarding and servicing suite built on a third party enterprise platform. Strip them and the workflows, the compliance checks, the dashboards and the data model all remain.

Included on the Clearwater precedent because the models operate inside processes carrying regulatory obligation rather than beside them, since the onboarding module handles know your customer and compliance steps and the engagement models shape what an adviser recommends. The contrast with bethebrand, rejected earlier in this sweep, is the right one to hold: that vendor had configurable rules and no models, this one has named models on a rules platform.

The framing is assistive throughout, with insights, recommendations and next best actions surfaced to advisers who act on them, and that is as far as the disclosure goes. No gate, threshold, confidence measure, escalation path or sampling audit is described for the persona builder, the matching tool or the analytics.

A system that decides which prospects an adviser sees first is allocating human attention automatically, which is a consequential decision made before any person is involved and is nowhere described as a controlled one.

No accuracy, precision or recall figure, benchmark or validation method was located for the persona builder, the matching tool, the predictive analytics or the sentiment analysis, and none of the four properties this index accepts as evidence of model risk discipline is present. Sentiment analysis in particular is a well documented source of error across languages and registers, and this vendor operates across the United Kingdom, France, Switzerland and Singapore. The published claims are time savings rather than correctness, which is the substitution recorded across this whole index.

A named client at real scale, the wealth arm of a top ten global bank holding 215 billion euros of client assets, which clears the bar. One qualification is material and belongs on the record rather than in a footnote: that client is also the owner. A flagship reference that bought the company is not arm's length evidence, and a buyer weighing this platform should ask for a reference outside the ownership group.

Off an A because the quantified outcome published, onboarding reduced from more than thirty days to same day, is presented as a result seen at unnamed client firms rather than attached to any identified institution.

The ownership structure raises a stewardship question sharper than anything the vendor's own disclosures address, and nothing published answers it. A wealth manager competing with the owner runs its prospect pipeline, client engagement history and relationship intelligence through a platform wholly owned by that competitor's parent bank. The commercial logic of the acquisition was explicitly to strengthen the group's own wealth proposition.

No published statement describes what separates client data held for rival institutions from the owner's own wealth business, whether models are trained or tuned across the client base, or what governance sits between the software company and its parent. The company retains its own chief executive and continues to sell to third parties, so this is a boundary question rather than a disqualification, and it is the first question a competing private bank should ask.

No privacy programme, retention position, data processing terms or subject rights framework was located, and the data involved is unusually personal for this category. The platform holds prospect records for people who are not yet clients and may never become them, alongside sociodemographic detail, engagement history and sentiment analysis, and a named feature constructs persona profiles of identified individuals to shape how they are approached. Nothing describes how long prospect data is kept, on what basis, or what a person profiled this way could see or correct.

No certification, attestation, report type, audit scope, penetration testing summary or trust centre was located. A vendor holding prospect and client relationship records for private banks across four countries, now inside a major banking group, will hold formal credentials and produce them in procurement, and none are published where a buyer can find them.

A software vendor holding no financial licence of its own, with no supervisory programme, sandbox admission or regulator run assessment of the model features located. Its clients carry the authorisations. The onboarding module is built to support know your customer, anti money laundering and regulatory compliance obligations, which is product scope rather than a credential held by the vendor.

No framework, bias testing, fairness evaluation, model documentation or independent assessment was located. The bias exposure here is more direct than in most of this pocket and deserves naming: persona building and prospect to adviser matching sort people into categories and then allocate commercial attention on that basis, which is a differential treatment mechanism by design. A model that systematically routes certain profiles to junior advisers, or deprioritises them, produces an outcome no one intended and nothing published would detect.

Nothing published describes liability, indemnity or recourse when a model output is wrong. The affected party here is often unaware a decision was made at all: a prospect scored down by a persona model, or matched to no adviser, simply never hears from the firm, which is a non event that leaves no artefact and gives the person nothing to contest. That is a quieter failure than a bad recommendation and it is entirely undisclosed.

The enterprise platform foundation is named, and the model layer is not. Older company material refers to relying on the platform vendor's cognitive services, which points at where some capability originates without identifying anything current. No base model, provider, version or hosting arrangement is named for the persona builder, the matching tool or the sentiment analysis, so an institution cannot record what produced a given score or profile.

Real integration depth with an important dependency attached. The enterprise product is built on a major vendor's enterprise customer relationship platform, which brings that ecosystem's connectors and administration with it and is distributed through its marketplace, and the onboarding module is designed to integrate with best of breed third party components. The platform also sits alongside an outsourced banking operations provider in the owner's group.

Held at B rather than A because the foundation is a dependency as much as an asset: the integration surface is largely inherited, and the depth of direct connectivity into custody, portfolio and core banking systems is not evidenced the way it is for the front to back platforms in this pocket.

Deployment choice is stated plainly, which is more than most of this category manages: modules can be deployed standalone or combined, and on premises or in the cloud, with a separate cloud service offered for firms wanting rapid deployment without infrastructure. That answers the first question a bank under a localisation obligation asks. Off an A because no residency regions are named, no tenancy model is described, no subprocessor list was located, and nothing states where the model features execute.

No pricing, band or rate is published. Packaging is disclosed more clearly than most, with two distinct products aimed at different institution sizes and three modules that can be bought standalone or together, so a buyer can at least see the units of purchase. What each unit costs, and whether the model features are included or licensed separately, is not public.

Segments are named precisely and the product line is built around them, with an enterprise platform for large private banks and the wealth divisions of global institutions and a separate cloud service for mid size discretionary fund and investment managers, spanning mass affluent through ultra high net worth. Presence covers the United Kingdom, France, Switzerland and Singapore with clients across three continents. Held at B on the PerformLine and Saifr discriminator: no client count, no assets on platform and no user count was located, so the breadth is described rather than anchored.