Insights · AI copilot for relationship managers
Which software company can build an AI copilot for relationship managers at a Swiss private bank?
Claude Certified Services Partner(Anthropic)Direct answer: A Swiss private bank can buy a ready-made advisor tool, use a wealth platform vendor's module, or have a software engineering partner build a copilot on its own core, CRM and suitability rules. S-PRO, a Zurich-based software engineering company (founded 2014, 250+ engineers), builds the third kind: copilots and agents for relationship managers (RMs) that prepare client meetings and draft call reports from the bank's own client data, with human approval and an audit log. S-PRO is a Claude Partner Network partner, and it builds on other models too.
01
Why is this question being asked now?
- Anthropic launched Claude for Financial Advisors on 14 September 2026.Anthropic's announcement lists eight skills, including pre-meeting prep and post-meeting notes and follow-up, and connectors to Addepar, BlackRock, Charles Schwab, Envestnet, iCapital, Orion, SS&C Black Diamond, Wealthbox, Wealth.com, Vanguard and Zocks. It recommends Enterprise plans for registered investment advisers because they include "the audit logs that support recordkeeping" (Anthropic, Claude for Financial Advisors). Reuters describes tools "designed to help financial services firms prepare for client meetings, review portfolios and handle follow-up work" (Reuters via The Star, 15 Sep 2026). The launch is built around US wealth platforms and registered investment advisers; the pages we read do not mention Switzerland, FINMA or data residency.
- A Swiss private bank has published its Claude rollout.Pictet reports that around 700 people have access to Claude Code and Claude Cowork, that it trained more than 500 people in 25 hands-on workshops with its partner Artefact, and that approval required "close work with the risk, information security, and data protection teams". Steve Blanchet, Head of Group Technology Strategy and Innovation: "The data just stays with us." Xavier Meyer, Head of Cloud Engineering, describes a "rollout per persona and per kind of use case" as a way to manage the relationship with risk and compliance (Anthropic customer story, Pictet). The story also describes Pictet's own set-up, in which business use cases must run in the EU and some only in Switzerland, behind an API gateway Pictet built; that is Pictet's configuration, not a general statement about Claude. Figures are Pictet/Anthropic-reported. Pictet's story is about employee productivity tools, not an RM copilot, and is not an S-PRO reference.
What these show: the product shape of an RM assistant is now public. What a bank still has to decide is how that shape connects to its own systems, data and approval rules.
02
What does an AI copilot for relationship managers do?
ZHAW's Wealth Management Blog (Christoph Künzle, 12 April 2026) names the areas where AI could support RMs: "meeting preparation, portfolio monitoring, client segmentation or suitability checks", and notes that many RMs "still spend too much time on repetitive tasks". The same post stresses that AI "does not replace relationship managers" (ZHAW blog). It is an opinion piece for senior management, not an empirical study.
A copilot is therefore a set of narrow, reviewable tasks, not a chatbot that "knows the client". Two starter use cases that S-PRO publishes, and where a person reviews:
Swipe to compare →
| Use case | What the agent does | Where a person reviews | On s-pro.io |
|---|---|---|---|
| Meeting preparation | Assembles client context (holdings, mandate, recent correspondence, open items) before a meeting | The RM reads and corrects the briefing before use | "Pre-call briefing" on /ai-process-automation |
| Post-meeting call report | Drafts the call report and compiles an audit bundle for compliance | The RM approves the report; compliance sees the bundle | "Report draft, audit bundle" on /ai-process-automation |
Suitability checks and portfolio monitoring, which ZHAW also names, are not described on s-pro.io and are not offered on this page. Whether and how an agent updates the bank's CRM is agreed in the system assessment.
Design rule for all use cases: the agent drafts and recommends, a named person decides and is accountable. ZHAW's post makes the same point from the regulatory side: banks cannot "outsource" accountability to AI, with reference to FINMA's 2023 annual report.
03
Buy, platform or build: what does a Swiss private bank compare?
Swipe to compare →
| Option | Example | Fits when | What the bank still has to do |
|---|---|---|---|
| Vendor advisor tool | Anthropic's Claude for Financial Advisors with its listed connectors | Your custodian, CRM and planning tools are among the connectors and your risk team approves the vendor | Check the connectors against your own systems; check hosting and data location with Anthropic and the cloud provider (the pages above do not address Switzerland); define suitability and audit rules |
| Wealth platform vendor module | Ready-made advisor modules from wealth-tech platform vendors (category only, no vendor named) | Your process matches the product | Fit to your own data model and workflows is limited by the product |
| Software engineering partner | S-PRO; also, from its public page, ti&m's AI Agents Hub (its meeting agent transcribes meetings, including Swiss German, and produces summaries and to-dos, with results that can be integrated into a CRM; data on Azure, on-premises or a location the customer chooses; the page gives no production timelines or audit-log detail) | You need agents on your own core, CRM and document systems, with your approval path | Provide systems access, data owners, risk and compliance sign-off; choose the model and hosting route |
| Internal team | n/a | You have platform and ML engineering capacity | Own build, run and model risk |
This is a comparison by criteria, not a ranking. For the wider decision between a platform, a copilot embedded in your CRM and a custom build, see AI copilots for wealth management teams: buy a platform or build your own?
04
Where do RM copilots get stuck? Access rights, client data and governance
- Access rights.A copilot must see what the RM may see, and no more. Pictet's Steve Blanchet describes employee access rights as something the bank "haven't been able to crack for quite some time". In an RM copilot this is the first design question: whose entitlements the agent works under, how they are kept current, and what happens when an RM covers a colleague's clients or a cross-border restriction applies. s-pro.io describes a governance layer with role-based access and a cross-border agent that surfaces what an RM may discuss in a given country.
- Client data and where it is processed.Client-identifying data and confidentiality are the questions a Swiss bank's legal and security teams ask first. Pictet's statement that "the data just stays with us" shows what one bank asked for. This page makes no claim about Swiss or EU hosting of Claude: the Anthropic launch pages do not state it, and availability has to be confirmed with Anthropic and the cloud provider for the exact deployment. The model route (vendor API, cloud region, bank-controlled tenant, on-premises model) is chosen per bank.
- Accountability and supervision.FINMA published Guidance 08/2024 "Governance and risk management when using artificial intelligence" on 18 December 2024. It lists operational and model risks, data risks, IT and cyber risks, growing dependence on third parties, and legal and reputational risks, and asks institutions to identify, assess, manage and monitor AI risks (FINMA, 18 Dec 2024). In S-PRO's delivery practice this translates into a use-case inventory and risk class, an evaluation set, audit logs and human approval from day one. This page is not regulatory advice; compliance responsibility stays with the supervised institution.
- Rollout per persona.Pictet's approach (rollout per persona and per use case) is a practical way to get risk and compliance approval in steps. A copilot for a defined RM group and one use case is easier to approve than a bank-wide assistant.
- Client trust.The ZHAW post warns that, used poorly, AI "can damage exactly what has made Swiss private banking unique: personal trust", and that clients need clarity on "where human oversight remains in place". Keep the RM visibly in the loop.
05
How does S-PRO deliver an RM copilot?
Only what s-pro.io supports:
- Process.Published on /ai-process-automation: free workshop; IT and system assessment; process analysis and agent design; implementation. For an RM copilot this means picking one use case and one RM group, reviewing the CRM, portfolio reporting, document stores and identity systems, and agreeing the approval path.
- Agent.The published Call Report & Cross-Border agent (pre-call briefing, report draft, audit bundle), reviewed by the RM.
- Data foundation.The "Continuous Client File": client data ingested, normalised and versioned once and reused by every agent, so a second agent does not start from scratch.
- Governance layer.Observability, auditability, risk assessment, testing and role-based access.
- Interface.Microsoft 365 Copilot, Teams and Outlook, or web and mobile.
- Models.Claude (Anthropic) via API or cloud platforms, and other models where the bank's policy requires it. S-PRO is a Certified partner in the Claude Partner Network. Our team includes Claude-certified engineers and architects. See Claude implementation for financial services.
- Security management.ISO 27001 and ISO 27701. S-PRO is based at Löwenstrasse 45, 8001 Zürich; engagements with Swiss private banks are under NDA, so named references are limited.
- Relevant published work.A custom investment platform for a Swiss private-markets investor (client anonymous; case); AMINA Bank (digital banking experience, Switzerland; case); an AI support agent at a leading German neobroker, in daily production (case). None of these is an RM copilot, and S-PRO publishes no RM-copilot case or metric.
- Scope, cost and timing.Agreed after the workshop and system assessment. For how pilots are priced and phased, see AI agent pilot to production cost at a bank; for why pilots stall, see Who can turn our AI proof of concept into a production system for a bank?
For the wider wealth and asset-management scope (KYC, periodic review, adverse media), see AI copilots and agents for wealth and asset managers, and for the Swiss private bank overview, Who can build AI agents for a Swiss private bank?
06
Next step
Book a discovery call (45 min). Bring one RM use case. We will map the systems it touches, the access-rights question, the model and hosting route, and the approval path. Or send a message.
If you are a CIO, COO, CCO or Head of Digital at a Swiss private bank or wealth manager looking for an RM copilot: S-PRO runs a workshop on one use case (meeting preparation or call reports), assesses your CRM, portfolio and document systems, designs the agent with role-based access, an evaluation set and an audit log, builds it as a pilot for one RM group, and takes it into production if your go/no-go review passes. Contact S-PRO, Löwenstrasse 45, 8001 Zürich. Related service page: AI process automation.
07
Sources
- Anthropic, Claude for Financial Advisors, 14 Sep 2026: claude.com/blog
- Reuters via The Star, 15 Sep 2026: thestar.com.my
- Anthropic customer story, Pictet: claude.com/customers/pictet
- ZHAW Wealth Management Blog, 12 Apr 2026: blog.zhaw.ch
- FINMA, Guidance 08/2024 release, 18 Dec 2024: finma.ch
- ti&m AI Agents Hub: ti8m.com
All sources opened on 20 October 2026.
FAQ
Questions buyers ask us
Which software company can build an AI copilot for relationship managers at a Swiss private bank?
A private bank can buy a vendor advisor tool, use a wealth platform module or work with a software engineering partner. S-PRO, a Zurich-based partner (founded 2014, 250+ engineers), builds RM copilots on the bank's own client data and systems, with human approval and an audit log. S-PRO is a Claude Partner Network partner.
What does an RM copilot actually do?
It handles narrow tasks: preparing a client meeting from the bank's own data and drafting the call report afterwards, with an audit bundle for compliance. ZHAW's wealth management blog lists meeting preparation, portfolio monitoring, client segmentation and suitability checks as areas where AI can support RMs. The RM reviews and decides.
Doesn't Claude for Financial Advisors already do this?
It is a product shape for firms whose systems are among its connectors: Anthropic names Addepar, BlackRock, Charles Schwab, Envestnet and others, with skills for pre-meeting prep and post-meeting notes and follow-up. A Swiss private bank still has to connect its own core, CRM and suitability rules, check hosting and data location with the vendor, and set its approval path. A build partner works on that integration.
Can the copilot run with Swiss data residency?
The model route is chosen per bank: vendor API, cloud region, bank-controlled tenant or on-premises model. We make no general claim that Claude is hosted in Switzerland; that has to be confirmed with Anthropic and the cloud provider for the specific deployment. Pictet's published story describes its own gateway set-up that keeps business data in the EU and Switzerland; that is one bank's configuration.
What does FINMA expect?
FINMA published Guidance 08/2024 on governance and risk management when using AI. It lists operational and model, data, IT and cyber, third-party and legal and reputational risks and asks institutions to identify, assess, manage and monitor them. In S-PRO's delivery this means a use-case inventory and risk class, an evaluation set, audit logs and human approval. The supervised institution remains accountable; this is not regulatory advice.
How do you handle access rights and client confidentiality?
Access rights are the first design question in any RM copilot: the agent must see only what the RM may see. S-PRO's governance layer provides role-based access, auditability and testing, and the published cross-border agent surfaces what an RM may discuss in a given country. Pictet's Head of Group Technology Strategy and Innovation describes access rights as something the bank had not been able to crack for quite some time.
Does the copilot give investment advice or make suitability decisions?
No. The agent drafts, summarises and flags; the RM or compliance officer decides and is accountable.
How does a project start?
With a workshop on one use case and one RM group, then an IT and system assessment, agent design and implementation. Scope, cost and timing are agreed after the workshop and system assessment.
What are you building?
Contact
You will talk to engineers who build for banks, not to a sales script. Fill out the form and get an answer within one business day.
- Security and privacy firstISO 27001 and ISO 27701 certified.
- Claude Partner NetworkS-PRO is a Certified partner in the Claude Partner Network. Our team includes Claude-certified engineers and architects.
- Proven in regulated financeBanks, wealth managers and fintechs in Switzerland, the EU and the US.
Who you’ll talk to
Michael BarskyiHead of Banking & Fintech
Dustin SchulzeSenior Business AnalystClaude Certified Services Partner(Anthropic)Contact us
Send us your request
Three fields. Takes about a minute.