AI readiness goes beyond PMS features. Discover why trusted investment data, enterprise context and flexible architecture are key to wealth management AI.
Recently, Yann Thiel of Julius Baer published an excellent article through the Swiss Association of Wealth Managers (SAM) arguing that AI readiness is becoming the next battleground among Portfolio Management System (PMS) providers.
Drawing on discussions with around 20 PMS vendors, he observed that most providers now cover approximately 90% of the traditional functionality required by an External Asset Manager (EAM), while only a small number have deployed AI capabilities that are genuinely operational today.
This research confirms what many in the market have been observing for some time.
That, as workflow and process functionality converges within software platforms, AI is increasingly becoming a selection criterion. Not because firms expect AI to replace their people, but because they expect it to improve productivity across the entire business.
At Point, we agree entirely. However, we would argue that AI readiness is ultimately an operating model challenge rather than PMS-centric challenge. AI is accelerating the need for firms to rethink how data, technology and workflows interact across the enterprise..
The premise that PMS functionality is converging is hard to dispute. Portfolio management, reporting, compliance workflows, rebalancing and onboarding are increasingly mature capabilities across the market.
Yet most EAMs don't operate on a single platform.
Their operating model typically spans a collection of systems, including but not limited, to:
Why?
Because no single provider delivers 100% of the capabilities required to support the full client lifecycle, and the typical EAM operating model has grown organically, acquiring capabilities over time, resulting in a patchwork of disconnected solutions.
The persistence of spreadsheets across wealth management is perhaps the clearest evidence of this. They survive not because advisers enjoy using them, but because they provide flexibility that highly structured siloed applications, designed to automate process, often cannot.
As a result, while a PMS may provide 90% of required capabilities (though in reality it is probably less), the actual operating model is distributed across multiple applications and processes.
Which creates a familiar challenge: fragmented data and fragmented context.
Historically, PMS platforms have been designed around workflow automation.
Their purpose was to standardise processes, increase efficiency, improve control and support regulatory compliance - and they do this very well. But, they were never designed to be open, flexible, enterprise-wide investment data platforms.
Increasingly, we're seeing some providers introduce dedicated analytics, data and intelligence capabilities alongside their traditional PMS offerings. This reflects a growing recognition that process automation and data exploitation are not the same thing.
The distinction matters.
The system owned the workflow, but that model is now beginning to change.
Agentic AI introduces a fundamentally different approach.
Instead of users navigating applications and triggering workflows, AI agents will increasingly orchestrate workflows on behalf of users.
Rather than:
Users will simply ask:
"Prepare review packs for every client with more than CHF 5 million AUM, identify suitability reviews due in the next 90 days and generate draft adviser briefing notes."
The workflow becomes dynamic as the ‘intelligence’ moves above the application layer. Which in turn has profound implications for traditional software as Agentic AI assumes many of those workflow responsibilities, not necessarily by replacing underlying systems, but by orchestrating them.
One of the most important observations in Yann's article is that AI quality is heavily dependent on data quality. Incomplete records, inconsistent classifications and poor governance will inevitably lead to poor AI outcomes.
We would extend that argument further to include context quality.
Today's typical wealth management operating model is frequently typified by fragmented data. Humans bridge those gaps remarkably well. They understand relationships between portfolios, clients, products, interactions, documents and business processes. People understand the context.
In contrast, AI operating inside any single solution doesn’t. An AI assistant embedded inside a PMS can only access the data available within that PMS. An AI assistant embedded inside a CRM can only access CRM context.
Neither has a complete picture of the client or the organisation, and without context, intelligence (whether ‘artificial’, ‘super’ or just ‘human’!) quickly breaks down.
As wealth management businesses move towards AI-enabled operating models, we believe two new architectural layers will emerge.
1. The Data Orchestration Layer
The first is a dedicated data foundation responsible for:
This layer combines data from PMS platforms, CRMs, custodians, onboarding tools and other enterprise systems.
Crucially, it provides deterministic analytics. Performance calculations, exposures, classifications, risk metrics and client analytics should be calculated once, governed centrally and trusted across the organisation.
Put simply: Don't let AI do your maths.
AI should consume trusted calculations, not invent them. The role of the investment data layer is to create a consistent, contextual and AI-ready representation of the business.
2. The AI Orchestration Layer
The second is an AI orchestration layer operating above both data and applications.
Its role is to:
Instead of embedding intelligence inside individual silos, intelligence becomes portable and enterprise-wide. So, the applications and individual solutions remain important.
But they become participants in a broader operating model rather than the centre of it.
This shift also introduces a strategic question that many firms are only beginning to consider: Who owns the intelligence?
Historically, firms have competed through client relationships, investment expertise and operational efficiency.
In an AI-enabled future, competitive advantage will increasingly depend on an organisation's ability to assemble, govern and exploit its own data.
Here the real asset isn’t the application or platform, rather the contextual data layer.
A wealth manager that owns its investment data foundation can:
If that intelligence remains trapped inside individual software platforms, firms risk limiting their future flexibility.
There is another reason to focus on data rather than AI itself.
Nobody knows which AI model will be leading six months from now, let alone five years from now.
Today's market leaders may not remain tomorrow's leaders.
New models will emerge. Costs will change. Regulations will evolve. Geopolitical considerations may influence technology choices. Resilience and business continuity will become increasingly important factors.
Firms may wish to use:
The ability to make those choices depends on architectural flexibility.
If data is tightly coupled to a specific AI provider or application, organisations risk creating a new form of vendor lock-in. By separating the data foundation from the AI layer, firms retain the freedom to adopt whichever technologies create the greatest value.
In an uncertain world, flexibility will prove to be one of the most important AI capabilities of all.
Yann Thiel's excellent article makes a strong case that AI readiness is becoming an increasingly important consideration when selecting a PMS. We agree entirely, but think the implication is even broader.
AI readiness is an operating model design consideration.
It shouldn’t be measured by whether a specific platform has embedded a chatbot or an intelligent assistant or an MCP. Rather, by whether an organisation can provide orchestrated AI with trusted, reconciled, contextual investment data and allow that intelligence to operate across its entire business.
As Agentic AI assumes more workflow and process responsibilities, value will shift away from individual applications and towards the data and AI foundations that connect them.
In the coming years, the strategic asset may not be the PMS, the CRM or even the AI model itself - it will be the investment data foundation and wider operating model.