What does AI for independent wealth managers actually cost? Most writing about AI tools starts with capabilities and ends with governance. This piece, however, starts with money. Our earlier article, AI Tools for Wealth Managers: A Dated Snapshot, Not a Verdict, compares the leading assistants on what they can do and what their consumer tiers cost a single user. This article, in contrast, answers the other procurement question: what business and enterprise plans cost per seat, what the industry systems cost, and which contract details matter more than the sticker price. The two pieces are companions, not competitors.
Every figure below is a published list price, checked on 17 August 2026. Where a vendor publishes no price, the table says so; in other words, nothing in this article is an estimate.
Why AI pricing looks different for an independent wealth manager
A large bank does not pay list prices; it negotiates enterprise agreements with volume discounts, custom data terms and dedicated support. A three-person independent wealth manager, by contrast, buys AI from the public price list with a company card. That changes three things. First, minimum seat counts and plan floors matter. Second, the subscription itself is almost irrelevant next to the partner time it takes to introduce a tool properly. Third, the small firm carries the same regulatory duties as the large one, with none of the negotiating power to reshape a vendor’s data terms. More on that structural difference: Banks vs EAMs.
Level 1: what AI assistants cost an independent wealth manager
These are the prices most independent wealth managers will actually pay for AI in 2026, because most firms meet it through a general assistant long before their portfolio management system grows an AI feature.
| Tool and plan | Per user/month, billed annually | Billed monthly | Minimum seats | Training on your content | EU/CH data residency |
|---|---|---|---|---|---|
| ChatGPT Business (OpenAI) | US$20 | US$25 | 2 | No training on business data by default | Enterprise plan only (ten regions) |
| Claude Team, standard seat (Anthropic) | US$20 | US$25 | 2 (up to 150) | None by default on Team and Enterprise | Not published — ask |
| Microsoft 365 Copilot Business add-on | From US$18 (list US$21) | US$25.20 | Requires Microsoft 365 licences | Not used for AI training, per Microsoft | EU Data Boundary available |
| Google Workspace with Gemini included | About US$7 (Starter), US$14 (Standard), US$22 (Plus) | Higher, plan-dependent | 1 | Not stated on the pricing page — check Workspace terms | Data regions on selected editions |
| Perplexity Enterprise Pro | US$40 (US$400 per year) | US$40 | Not published | No training on enterprise customer data, per Perplexity | Not published |
| Mistral Le Chat Team | US$24.99 (US$50 monthly minimum) | Not published | Effectively 2, via the minimum | Opt-out — training is on until switched off | EU-based vendor; enterprise deployment options |
| DeepSeek app | No paid seat plan published | — | — | No enterprise terms published | Not published |
| Alibaba Qwen app | No business seat plan published | — | — | No enterprise terms published | Not published |
Four notes on the assistant table
First, consumer tiers (ChatGPT Plus, Claude Pro, Perplexity Pro and similar) are deliberately absent; they are covered in the feature comparison, and client-related work does not belong on them in any case. Second, Mistral is the outlier on training: on its Free, Pro and Team tiers the pricing page lists model training with an opt-out, which means it is on until someone switches it off. Third, Google no longer sells workplace AI as an add-on at all — Gemini is bundled into Workspace editions, which makes its price hard to isolate but easy to budget. Fourth, the two Chinese rows are deliberately thin: DeepSeek and Qwen ship consumer apps but publish neither business seat plans nor enterprise data commitments — for client work, that is the answer. Where they genuinely compete, by the token, we give them a full treatment in a separate article. Sources: the published price pages of OpenAI, Anthropic, Microsoft, Google, Perplexity and Mistral, all checked on 17 August 2026.
The three columns that matter more than the price
Yet for an independent wealth manager in Switzerland, the interesting part of an AI price list is not the dollar column. Instead, it is the training default, the hosting region and the contract tier at which the enterprise commitments actually start. After all, a five-franc difference per seat is noise. Whether prompts are used for model training, whether data can be kept in the EU or Switzerland, and whether those promises apply at the tier you actually bought — that is the substance. If a plan does not give you a data processing agreement, audit logs and a training default of off, then treat it as a personal productivity tool, not client infrastructure.
Seat plans are only half of the price list. The same models are metered by the token through their APIs — the billing that applies when a portfolio management system or CRM calls AI in the background, and the level at which the Chinese providers DeepSeek and Qwen genuinely compete. Because that arithmetic deserves its own tables, we cover it in a separate token-pricing snapshot. For an independent wealth manager, the practical point of the AI token market is simple: the raw intelligence underneath every seat plan is priced in cents, not francs — what you pay for above that is packaging, controls and accountability.
Level 2: what the industry systems cost
The portfolio management and CRM layer of the wealth management technology stack increasingly ships with AI features — reconciliation checks, report drafting, meeting summaries. Here, however, the pricing culture is the opposite of the assistant market: almost nobody publishes numbers. We looked at the AI-capable vendors most visible among Swiss independent wealth managers; our separate overview of CRM and PMS systems in Swiss wealth management covers the functional side.
| System | Category | Published price |
|---|---|---|
| WealthArc | Portfolio management, Swiss EAM focus | CHF 16,000 per year licence with ten portfolios included; CHF 400 per additional portfolio; CHF 5,000 onboarding, waived on contracts over two years; 20% discount for VSV members |
| Assetmax (Profidata) | Portfolio management / CRM | On request |
| Expersoft PM1 | Portfolio management | On request |
| Allocare AMS | Portfolio management | On request |
| Etops | Platform and outsourced operations | On request |
| Evooq | Advisory technology | On request |
| Unique | Meeting and compliance AI | On request |
That column of on request is not laziness on our part; it is how this market sells — quote-based, usually after a demo. Indeed, WealthArc’s published list, verified on 17 August 2026, is the exception rather than the rule. In wealthtech, published pricing remains the exception, which makes WealthArc’s public list a useful anchor for budgeting a portfolio management system even if you end up buying elsewhere.
A worked example: 3, 10 and 25 staff
For instance, take an independent wealth manager that gives every employee one AI assistant seat, at list prices, billed annually:
- 3 staff: Microsoft 365 Copilot Business add-on about US$648 per year; ChatGPT Business or Claude Team about US$720; Perplexity Enterprise Pro about US$1,200.
- 10 staff: roughly US$2,160 (Copilot) to US$4,000 (Perplexity), with ChatGPT Business and Claude Team at US$2,400.
- 25 staff: roughly US$5,400 to US$10,000 per year — the scale at which enterprise negotiations, custom data terms included, start to make sense.
Set that against the industry layer: the one published PMS licence runs CHF 16,000 per year before portfolio surcharges. In fact, even at 25 seats, AI subscriptions cost less than the core system. The cheque that hurts is not the AI cheque — precisely why these decisions get taken casually, and why governance tends to lag the spend.
What the list price does not cover
- Introduction and training. To begin with, prompting habits, internal guidelines, a named owner. Budget days of partner time, not hours.
- The compliance review. Additionally, a data protection assessment, updated internal directives, client information where required.
- The outsourcing file. Furthermore, vendor due diligence, contract review and an exit scenario, documented in line with FINMA’s outsourcing expectations.
- The audit trail. Enterprise plans provide logs; nevertheless, someone still has to own, review and retain them.
- Switching costs. Snapshot pricing invites re-shopping every year; however, retraining a team is not free.
Equally, efficiency claims deserve their own scepticism — we have looked at them in Maximising Efficiency with AI in Wealth Management and at the broader Swiss wealthtech landscape.
A necessary warning on AI for independent wealth managers
None of the AI prices above buys an independent wealth manager an exemption from the rules of the profession.
- Client data and client names do not belong in consumer tiers of AI tools. If a plan trains on input by default or offers no data processing agreement, then it is not a place for client information.
- Likewise, professional secrecy, bank-client confidentiality and Swiss data protection law apply unchanged. A chat window is not an exception clause.
- Similarly, FINMA’s expectations on outsourcing and on AI governance — set out in its Guidance 08/2024 of 18 December 2024 — extend to AI services: inventory, risk classification, clear responsibilities, monitoring.
- Above all, language models produce confident errors. No investment, tax or legal text leaves the firm without human review.
- Finally, prices and features in this market change monthly. This article is a snapshot dated 17 August 2026; treat anything older than a quarter as history.
FAQ
Can an independent wealth manager use a private AI account like ChatGPT Plus for client work?
For drafting a marketing text that contains no client information, perhaps. In contrast, for anything touching client data the answer is no: consumer tiers lack the contractual commitments that business tiers carry. Moreover, those business tiers start at about US$20 per user per month, which removes the last economic excuse.
At which plan level do the enterprise data commitments begin?
It varies — and it is the most important line to check. For example, OpenAI and Anthropic switch off training by default from their business tiers upwards, while residency commitments sit higher — at OpenAI, on the Enterprise plan. Meanwhile, Microsoft ties its commitments to the Microsoft 365 stack, including the EU Data Boundary. Mistral, by contrast, requires an explicit opt-out below enterprise level.
Do these assistants replace a portfolio management system?
No. Instead, they sit beside it; the PMS and CRM layer has its own — largely unpublished — economics and its own selection criteria.
How reliable are these prices?
They matched the vendors’ public price pages on 17 August 2026. Expect drift within months rather than years; hence the date sits in the title rather than in a footnote.
Published list prices, checked on 17 August 2026. This article reflects personal views, simplifies where necessary and is not investment, legal or tax advice. Therefore, verify current prices and contract terms with the vendors before deciding anything.
