AI for independent wealth managers: 7 use cases that already work today

Seven use cases already work reliably at independent wealth managers today: writing meeting minutes, summarising research, producing client reporting, drafting client correspondence, making internal directives searchable, checking onboarding documents and preparing KYC dossiers. The difference between an experiment and productive use rarely lies in the tool. It lies in the order: firms that start with processes that consume a lot of time and require little sensitive data reach results within weeks.

The study «AI Use Cases and Crypto Assets in External Asset Management» by Lucerne University of Applied Sciences and Arts (HSLU) and Finup from July 2026 describes the starting point precisely: 49 percent of independent wealth managers in Switzerland already use AI in selected processes. Yet of the 65 reported applications, only 11 percent run productively; two thirds remain experiments (HSLU/Finup, July 2026). This article is written for owners and managing directors who know they should act but do not know where to start. It answers that in three parts: which cases pay off first, how to prioritise them correctly and what the first 30 days look like.

Which seven use cases already work today?

The short answer first, one sentence per case:

1. Meeting minutes: A structured, archivable advisory protocol emerges from the recording or your bullet points.
2. Research summaries: Research reports and market commentaries are condensed to one page, with sources.
3. Client reporting: Portfolio data is extracted from bank reports, consolidated and prepared as an understandable client report.
4. Client correspondence: The agent drafts emails and quarterly cover letters; you review and send.
5. Knowledge search: Internal directives and process documents become searchable in natural language, with a reference to the exact passage.
6. Onboarding documents: Incoming documents are read and checked for completeness against your checklist.
7. KYC and source-of-wealth dossiers: The agent creates dossier drafts from public sources and existing documents; the review stays with the human.

How do I prioritise the use cases correctly?

Two questions are enough to determine the order. First: how much time does the process cost today, every week, every month? Second: how sensitive is the data it touches? That produces a simple matrix:

	Low data sensitivity	High data sensitivity
Large time savings	Protocols, research, knowledge search	Client reporting, KYC dossiers, onboarding documents
Smaller time savings	Drafts for standard correspondence	Individual investment communication

Start top left. Client reporting sits on the right because of the asset data involved; once a protected environment is in place that the data never leaves, it can still be tackled early.

Top left holds the cases that return a lot of time and touch little sensitive data. Start there. Bottom right holds what you can safely postpone or deliberately leave with humans. This order is also my answer to the most common trap: many firms start with KYC precisely because that is where the pain is greatest. But KYC is the case with the highest data sensitivity and the most demanding data foundation. Starting there means months to a first result and a team that loses faith along the way. Starting with protocols or research puts numbers on the table within weeks, and after that the team asks for more on its own.

The seven use cases in detail

1. Meeting minutes under FIDLEG

What: From the meeting recording or your bullet points, the agent creates a structured protocol: the client's concerns, the proposals discussed, risks, agreements, next steps. You review the draft, amend and archive it.

Effort: Low. The agent needs your protocol template and access to the archive. No interface to banking systems required.

Data risk: Medium. The protocol contains client information but stays internal. With Swiss hosting and clear access rights, that is well controllable.

Example: In my projects, the effort for a clean advisory protocol drops by 90 to 95 percent. Twenty minutes of follow-up become two minutes of review. Across a year with several hundred client meetings, that is the largest single lever I know. As a by-product, it creates the documentation FIDLEG expects from financial service providers, as a side effect rather than extra work. The exact documentation duties depend on your activity; see the note at the end of this article. The full workflow, from consent through the fixed protocol structure to filing, is described in the article meeting minutes and FIDLEG documentation with AI.

2. Research summaries

What: The agent reads research reports, factsheets and market commentaries and summarises them on one page, with sources. In German if you wish, as a basis for the investment committee or for a client email.

Effort: Low. Public or subscribed documents are sufficient input. No client reference needed.

Data risk: Low. Nothing client-specific flows in.

Example: The study lists investment research among the most common AI use cases in the industry and at the same time among the biggest pain points. In daily life this looks like: a wealth manager with five people has the ten relevant reports condensed to half a page each, every morning. The committee reads for ten minutes instead of two hours.

3. Client reporting

What: The agent reads bank reports, extracts positions and transactions, consolidates across banks, currencies and products, and prepares the client report, including performance calculation.

Effort: Medium. It requires a connection to your custodian banks' formats. Once built, the process runs repeatably.

Data risk: High, but controllable. Asset data stays inside your protected environment and is not passed to public models.

Example: T&T International, a wealth manager with a very small team, handles high volumes of compliance work and reporting with an operations application. It reads bank reports, prepares the required documents including the compliance check, and shows the client in aggregated, understandable form what they hold where. The performance calculation follows best-in-class methods, and the reporting quality partly exceeds bank level. The time required has dropped by 80 percent. The details are in the T&T International case study. How the reporting is built there in detail and what other wealth managers can take from it is described in the article Automating client reporting: the T&T International case study.

4. Drafts for client correspondence

What: The agent drafts emails, quarterly cover letters and answers to standard questions. It takes tone and facts from your briefing. You review, adjust and send.

Effort: Low. A handful of templates and sample texts are enough to start.

Data risk: Medium. Client reference yes, but no transaction data needed.

Example: In the enterprise world, an advisor assistant introduced at one of the world's largest insurers sends occasion-based emails automatically, for instance for a birthday, and enriches the CRM with soft client data via its voice function. For a small firm, half of that is enough: the agent drafts, you send. That alone saves noticeable time, because the empty screen disappears.

5. Knowledge search across internal directives

What: Semantic search across your directives, compliance requirements and process documents. You ask in natural language; the agent answers with a reference to the right passage in the right document.

Effort: Low to medium. The documents need to be filed cleanly, nothing more.

Data risk: Low to medium. This is about internal documents, not client data.

Example: «What is our policy on accepting third-party funds?» Instead of twenty minutes of folder searching, the answer arrives in seconds, with the source passage. Especially in small teams, where knowledge sits in heads and email threads, this is a quiet but daily gain. A related variant is described in the article AI for family offices: making the entire document archive of a firm searchable.

6. Onboarding documents

What: The agent reads incoming documents, checks completeness against your checklist, lists missing documents and prepares the dossier for your review.

Effort: Medium to high. Every firm has its own forms and its own risk policy that must be mapped cleanly.

Data risk: High. ID documents, proof of wealth and tax documents are among the most sensitive material in the firm.

Example: Here I deliberately write from an industry perspective, not from a project of my own. From my work at additiv I know how large institutions rethink onboarding: in the discretionary portfolio management of a European neobank market leader, clients guide themselves through the process while the processing around them runs automated in the background. Translated for an independent wealth manager that means: the digital client experience does not come first, the completeness check in the background does. The client only notices that things move faster. The full process with its six steps and an honest assessment of where AI saves time and where it does not is described in the article client onboarding at wealth managers.

7. KYC and source-of-wealth dossiers

What: The agent creates KYC dossier drafts from public sources and existing documents: commercial register, media reports, your own notes. It structures the origin of assets as a draft. Review and responsibility stay with the human.

Effort: High. It needs clean data and clear rules, otherwise it creates more work instead of relief.

Data risk: High. This is where the largest lever and the greatest duty of care sit closest together.

Example: Again: the study lists KYC and AML checks among the most common fields of use and at the same time among the biggest pain points. That tension is exactly the point. What can be automated is the preparation, not the responsibility. That is why this case belongs at the end of the list, not the start. Firms that tackle it after two or three more successful cases bring the routine it requires. The full process, from reading the documents to source referencing with human checkpoints, is described in the article preparing KYC and source-of-wealth dossiers with AI.

What are the large institutions doing, and what is enough for five people?

What I take from projects with additiv: insurers and neobanks embed AI directly into their advisory processes. Three patterns stand out: hybrid advice, where AI supports the advisor in preparation and follow-up. Guided advisory applications that let advisors offer complex topics and holistic financial planning. And discretionary portfolio management with self-service onboarding that automates the processing around the client.

These platforms are multi-year projects with dedicated teams. For a firm of five to twenty people, rebuilding them is neither necessary nor sensible. What counts is the logic behind them: capture data cleanly once, prepare documents automatically, free the human for the advice. The seven use cases above are the weeks-long version of that. What insurers and neobanks build over years, a small team can use in weeks today, provided it starts in the right order.

What do the first 30 days look like?

Week 1: Choose one process. Pick a case from the top-left quadrant of the matrix. Make the goal measurable: minutes per protocol, reports per month, search effort per week. Without a target number there is no basis for a decision later.

Week 2: Clarify data and hosting. As soon as things get concrete, one question sits on top: where does my data live? The good answer is simple today: a protected Swiss cloud, hosted for example on Microsoft Azure in the Zurich North region. Client data stays in the country and cleanly separated from public models.

Week 3: The pilot runs in assisted mode. The agent creates drafts, your team reviews, nothing goes out without approval. Every action is logged with a timestamp and rationale. That is exactly the trail supervisors expect: in its Guidance 08/2024, FINMA sets out expectations for governance and risk management when using AI, namely that you know where AI runs, who is responsible and how errors surface (FINMA, December 2024).

Week 4: Measure and decide. Hit rate, time saved, error types. Then decide: scale, adjust or stop. Only after that comes the second process.

On budget: a lean pilot including setup sits in the mid five-figure range, with moderate monthly running costs on top. The detailed breakdown is in the article What does AI cost for financial service providers?.

From my practice: The largest effect rarely comes in the first process, but in the third. Reporting at T&T International cut the time required by 80 percent. The actual result, though, was different: once the team had experienced that an agent prepares reliably and the human only reviews, the next processes landed on the list by themselves. Trust in AI is not created by a presentation, but by measured quality over weeks.

Conclusion: your next step

Half the industry experiments, and of the reported applications only about one in ten runs productively. The difference is not technology but a decision: one process, one measurable goal, one accountable person. From the matrix above, choose the case that costs you the most time every week and touches the least sensitive data. That is your pilot.

If you would like to see what this looks like for your firm: on AI for wealth managers we show the use cases in detail and are happy to arrange a conversation. The full overview of AI in financial SMEs is in the 2026 guide, and the commentary Only 11 percent of AI applications at wealth managers run productively analyses why so many AI projects stall.

Note: We are not a law firm. All statements on legal and regulatory topics are subject to change and do not replace legal advice. For specific questions, please consult a lawyer.

FAQ

Is AI worthwhile for a team of three?

Yes, especially there. The smaller the team, the larger the relative gain per automated hour. What matters is starting with one clearly bounded process instead of a broad programme.

Which use cases should I not automate?

Investment decisions and individual advice belong in human responsibility. So does anything where the data flow is opaque or you do not control the sources. AI prepares, the human decides.

How do I handle data protection?

Three ground rules: Swiss hosting for client data, assisted mode with human approval, a complete log of every action. With that you answer the question «where does my data live?» cleanly, before it is even asked. What your team may and may not do with public tools like ChatGPT is covered in ChatGPT and client data: what Swiss wealth managers may and may not do.

What does getting started cost?

That depends on the chosen case: a first pilot including setup is typically feasible in the five-figure range, and the running costs stay small compared to a new hire. Details and the payback calculation are in the cost article.

Do I have to report my use of AI to FINMA?

That depends on your licence and supervisory regime. FINMA sets expectations for governance and risk management, not a general reporting duty for every use of AI. Clarify your specific case with a specialist; see the note above.