Most external asset managers (EAMs) rely on a Portfolio Management System (PMS) to consolidate assets across multiple custodians. During the summer of 2026, we met with PMS providers to assess which solutions best meet the diverse needs of EAMs.

We asked providers to demonstrate the capabilities required by a modern Swiss EAM: digital onboarding, CRM, portfolio construction, rebalancing, compliance workflows, reporting, and artificial intelligence (AI).

The findings were striking. Most providers now cover a substantial portion of the traditional functionality required by an EAM. As a result, functionality alone is no longer the primary differentiator. A new battleground is emerging: AI readiness.

AI is becoming a real selection criterion

Until recently, EAMs selected a PMS based primarily on custodian connectivity, reporting capabilities, portfolio management tools, compliance functionality, and pricing. These factors remain important, but the gap between providers is narrowing. Most platforms now cover the vast majority of an EAM’s core operational requirements. The next differentiator may therefore be AI.

This is not because AI is replacing portfolio managers or relationship managers, but because it can eliminate a significant amount of repetitive administrative work. Providers are increasingly using AI to support meeting preparation, document analysis, report generation, client onboarding, CRM updates, KYC processes, investment research, and data retrieval.

Beyond functionality, the key question is whether AI can materially improve advisor productivity and, ultimately, EAM profitability. As a result, EAMs evaluating a PMS should not only assess what the platform can do, but also how efficiently work can be performed within it.

From ChatGPT to embedded AI

Most financial professionals are familiar with ChatGPT, Claude, and similar tools. However, using these solutions with confidential client information remains challenging. This has accelerated demand for AI capabilities directly embedded within PMS platforms. The objective is simple: allow advisors to query portfolio-, CRM-, and operational data using natural language while keeping sensitive information within the PMS environment.

Instead of navigating predefined reports, users can simply ask: “Which clients have more than CHF 1 million invested, more than 10 per cent liquidity, and no documented suitability review during the last twelve months?” The ability to obtain such answers in seconds has the potential to transform advisor productivity by reducing their administrative workload and allowing advisors to spend more time with clients and portfolio management.

What we observed

Of the PMS providers with which we are currently in contact, a small number of providers demonstrated AI capabilities, as of August 2026, that could be classified as genuinely operational today. The most common use cases included:

• AI knowledge assistants explaining how to use the PMS
• ‘Chat with your data’ capabilities across portfolio and CRM information
• AI-powered report generation
• AI-assisted document analysis and parsing functions
• Meeting preparation agents
• Automated KYC and onboarding processes, including the extraction of client information and the pre-population of bank account opening forms
• Dashboard creation through natural language
• AI-assisted investment research and portfolio analysis

Some providers already allow users to upload a custom reporting template and automatically generate reports in the desired format. Others demonstrated document intelligence, enabling advisors to ask questions such as: “Who are the beneficiaries of the trust ABC?”

One provider demonstrated an AI assistant directly available within the end-client portal. Another showcased a fully in-house AI infrastructure running on its own GPU* environment and supported by a dedicated AI development team. Several providers are also building Model Context Protocol (MCP) layers to connect PMS data with external AI solutions.

Despite the growing interest in AI, the maturity gap between providers remains significant. Many are still presenting roadmaps, while only a handful have deployed solutions that can be used in daily operations. This raises broader questions about the level of investment that may be required across the industry as AI capabilities become increasingly expected by clients.

We also observed two distinct technology strategies. Most PMS providers with AI capabilities currently rely on specialist third-party firms such as Olymp, Dafin, Unplex, and others*. Only a few providers demonstrated solutions primarily based on open-source models running within their own infrastructure and data environments.

However, the direction of travel appears clear: as AI becomes a core component of the PMS, many providers are seeking greater independence and are gradually building proprietary AI capabilities on top of open-source models. Greater control over data, costs, and intellectual property appears to be a key driver behind this trend.

Data quality: The hidden prerequisite

While much attention is focused on large language models, the real differentiator may lie elsewhere: the quality of the underlying data.

An AI assistant is only as good as the information it can access. If client records are incomplete, classifications inconsistent, or CRM information poorly maintained, even the most sophisticated AI model will generate unreliable results.

This challenge will become even more important as PMS platforms move away from static dashboards towards conversational analytics. Instead of navigating predefined reports, advisors will increasingly build dashboards and analyses simply by asking questions.

Data governance, data quality, and integration may soon become more important competitive advantages than the choice of AI model itself. Questions around data residency, client confidentiality, and cybersecurity are becoming increasingly important evaluation criteria for EAMs adopting AI-enabled platforms.

AI innovation must not come at the expense of compliance

The race to introduce AI capability also raises an important question: how many providers are focusing on AI before mastering the fundamentals?

For Swiss EAMs, AML controls, KYC processes, suitability assessments, audit trails, and FINMA-compliant workflows remain non-negotiable.

The most impressive AI assistant creates limited value if the underlying compliance framework remains fragmented. In addition, firms will need robust controls to address AI hallucinations and ensure that AI-generated outputs remain explainable and auditable.

When evaluating PMS providers, EAMs should therefore ask two questions:

1. Does the platform solve today’s operational and regulatory requirements?

2. Is it building the AI capabilities that will create value tomorrow?

The race has only just begun

The PMS market can no longer be evaluated solely on traditional functionality. Increasingly, the differentiator is not whether a provider offers AI, but whether that AI is genuinely useful and embedded into everyday workflows.

The winners will not necessarily be those with the most powerful language model. They will be those that can combine trusted data, regulatory rigor, intelligent automation, and a sustainable economic model for both PMS providers and EAMs.

In a market where most PMS providers already cover the fundamentals, AI readiness may ultimately become the defining factor separating the leaders from the followers. The ultimate beneficiaries should be end clients, who can expect faster service, more personalised reporting, and better-informed interactions with their advisors.

Please note: GPU (Graphics Processing Unit) refers to specialist computing hardware used to train and run AI models. Olymp, Dafin, and Unplex are examples of specialist fintech providers that help financial institutions deploy AI capabilities such as document intelligence, workflow automation, AI assistants, and large-language-model integration. Examples are provided for illustrative purposes and do not constitute a recommendation.