Global signals. Swiss focus.

Which developments will change your business model over the next five years? We put eight megatrends in context with current figures from Switzerland and abroad and show what they mean for your institution’s revenues, costs and risks. The briefing ends with a prioritisation for your board: what to do now, and which strategic options to keep open.

What we cover

• Generative and agentic AI: costs, advice and the customer interface are shifting at the same time, and dependence on a few technology providers is growing.
• Pressure on efficiency and consolidation: cost ratios are rising despite record assets, and the number of institutions is falling. What does that mean for your independence?
• Cyber and third-party resilience: service providers, cloud and outsourcing become entry points and therefore a core risk the board is accountable for.
• Tokenisation and digital money: stablecoins, tokenised deposits and digital central bank money redistribute deposit, payment and custody business.
• Platform and neobank competition: app-based banks are winning hundreds of thousands of customers in Switzerland and seeking a banking licence of their own to move into core banking.
• Generational wealth transfer: how to secure the relationship with the heirs of your current clients.
• Instant payments and open finance: when payments arrive in seconds and third parties access account data through interfaces, the role of the main bank changes.
• Quantum-safe cryptography: a multi-year transition that needs to be planned and budgeted today.

We visit eight hotspots and flagship projects of the global financial industry, without you having to travel. The focus is the future of customer service. You see what already works elsewhere, with the figures behind it, and assess with us what fits Switzerland.

What we cover

• Silicon Valley and San Francisco: card networks, payment providers and AI labs are building the infrastructure through which AI agents shop and pay on customers’ behalf.
• Greater Bay Area (Hong Kong, Shenzhen): mass-market payments by AI agent, AI-native digital banks and a regulated framework for stablecoins and tokenised deposits.
• Singapore: DBS measures the value of its AI applications and reports it at group level. A template for AI governance in the boardroom.
• London: neobanks with millions of customers steer their service by resolution times and referrals.
• Stockholm: Klarna as a lesson: an AI assistant was meant to replace hundreds of service jobs, followed by a return to more human quality.
• São Paulo: with Pix, the central bank’s instant payment became Brazil’s most used means of payment, and Nubank a digital bank for the wider population.
• Mumbai and Bengaluru: India’s public infrastructure for identity and payments turns payments into a commodity. Banks have to differentiate in other ways.
• Dubai: the DIFC financial centre is growing fast among family offices and wealth managers and is becoming a direct competitor to Switzerland as a financial centre.

Trips 2027: In 2027 we plan small-group trips to Silicon Valley and the Greater Bay Area, including Hong Kong and Shenzhen. On site, you experience what the briefing shows: payments by AI agents, AI-native banks and the regulation of digital money. We will share the programme, dates and terms with interested participants early. Register your interest

What can generative AI do today, and where are its limits? We explain the technology so that your board can put projects in context, challenge results critically and ask the right questions.

What we cover

• What works today: language and reasoning models, software development, operating software. And why these capabilities change within months, not years.
• The economics of AI: the cost per request is falling sharply. That changes make-or-buy decisions and makes long licence commitments risky.
• Where AI is unreliable: invented content, and top performance alongside simple mistakes. Controls therefore belong in the process, not at its end.
• Legal and liability framework: the Data Protection Act, FINMA’s expectations and liability for what a chatbot tells customers.
• A grid for your board: assistance or decision, data sensitivity, customer impact, consequences of errors and human control. It lets you place any AI project.

Customer communication is becoming more personal, faster and increasingly mediated by AI. More and more customers ask an AI first and their bank second. We show how customer access and brand management are changing and where your institution should act.

What we cover

• From search to AI answer: if you do not appear in the answers of ChatGPT, Gemini or Google AI Mode, you lose the first contact.
• AI visibility as a brand metric: how to measure it, what role ratings and third-party sources play, and how to respond to false AI statements about your institution.
• The next generation: heirs get their information from AI and finfluencers. How do you keep the client relationship through the generational change?
• Hyper-personalisation: data foundations, consent and the point at which personalisation costs trust.
• Guardrails: advertising under FinSA, working with finfluencers and approval processes for content created by AI.
• AI agents as customers: when agents compare offers and sign up, product and price information has to be machine-readable.

AI assistants such as Microsoft Copilot and new analytics tools are changing planning, reporting and steering. We show what these tools deliver today, which efficiency gains are realistic and which requirements for data, systems and governance must be met.

What we cover

• Reality check: use is widespread, measurable benefit often still small. Ask about impact, not the number of pilots.
• What copilots do today: reconciliations, variance analysis, closing, receivables and planning, illustrated with Microsoft 365 Copilot and SAP Joule.
• The right metrics: time to close, forecast accuracy and error rates instead of estimated time savings.
• Prerequisites: data quality, integration with the ERP system, data residency in Switzerland and outsourcing questions.
• Governance and auditability: figures that flow into the financial close and board papers need clear accountability and human control.
• Buy or build: and how the role, skills and size of the controlling function change.

A bank’s board regularly approves the strategies for ICT, cyber risk, critical data and business continuity, and monitors compliance with them. We prepare these topics so that your board can assess them. For institutions with an EU connection, we cover what the DORA regulation requires of the management body.

What we cover

• The threat picture for Swiss finance: attacks via service providers, insiders and AI-assisted methods, and which of them affect your institution.
• The board’s duties: approving and overseeing the strategies under FINMA Circular 2023/1. For groups with entities in the EU, DORA applies as well, and in Germany also the German Banking Act (KWG).
• Competence on the board: a member with an IT background, a committee, external advice or training for everyone. The aim is not a technician on the board but the ability to challenge management critically.
• Reporting that enables decisions: a few key figures, clear thresholds for escalation to the board and the results of attack simulations.
• Crisis readiness: how a board rehearses a cyber crisis, with decisions under time pressure instead of a presentation.

Critical functions and tolerances for disruption are approved by the board. FINMA expects these tolerances to reflect the board’s stance towards shocks. We help you clarify that stance before you approve the numbers.

What we cover

• What applies since 1 January 2026: the transitional period of FINMA Circular 2023/1 has ended. Operational resilience must be ensured.
• Deriving tolerances properly: the board first defines what harm to customers, the market and the institution is intolerable. Only then comes the number, and not from what IT can restore today.
• Defining critical functions correctly: processes and resources are not functions. Each function needs an end-to-end view from front to back office, covering people, systems, data and suppliers.
• Third parties and concentration: exit strategies, subcontractors and testing rights with cloud and core banking providers.
• Severe but plausible scenarios: including those without an attacker, such as a faulty update, the failure of a cloud provider or an outage of the core banking system.

FINMA expects institutions that use AI to align their governance and risk management accordingly. Where applications are numerous or material, it reviews inventory, responsibilities, testing, documentation and training, among other things. We show which questions the board should ask and how it can recognise good AI governance.

What we cover

• Accountability and organisation: full board or committee, and how AI oversight is anchored in the organisational regulations.
• Inventory and risk classes: classification, testing, documentation and explainability under FINMA Guidance 08/2024.
• Dependence on providers: contracts, exit strategies and concentration risk with a few AI and cloud providers.
• Reporting to the board: a minimum set covering inventory, risk classes, incidents, benefits and training status.
• Regulatory radar: the Data Protection Act, Switzerland’s planned approach and, for EU business, the AI Act’s high-risk obligations from December 2027.

AI agents increasingly act on their own: they analyse, prepare decisions and trigger tasks. We show where this is already happening in banking, what opportunities and risks arise, and where human oversight remains essential.

What we cover

• From assistant to agent: what matters is the level of autonomy: prepare, recommend or execute. And who approves the next step?
• State of practice: assistants are widely deployed at large banks. Real agents work in tightly scoped processes such as lending, KYC and compliance, with humans in control.
• Payments by AI agents: card networks and technology firms are setting the standards. Who will own the customer interface?
• Control architecture: identity and permissions per agent, step limits, logging, kill switches and separate test and production environments.
• A realistic business case: proceed in stages, set measurable goals and secure exit options from providers.

Powerful quantum computers could break today’s encryption. FINMA recommends a strategy adopted by the board and a roadmap for post-quantum cryptography by mid-2027 at the latest. We explain what your board needs to know to adopt that strategy.

What we cover

• The picture without hype: what is established, what is vendor roadmap and why the exact timing is secondary for your decision.
• “Harvest now, decrypt later”: data intercepted today can be decrypted later. Which of your data must remain confidential for longer than the transition will take?
• Deadlines set the pace: FINMA, the EU, the UK and the US are setting timelines that effectively pull suppliers and financial market infrastructures along.
• Crypto inventory and crypto agility: where asymmetric encryption is used, from interfaces to hardware security modules, and how quickly it can be replaced.
• What belongs in the strategy: accountability in the executive team, a multi-year budget, milestones and reporting to the board.