adopting teams
AI and automation solutions in active use
AI Product & Platform Leader · Zürich
Building AI capabilities that financial institutions can actually use.
I work across wealth management, AI products, automation and enterprise delivery—turning promising technology into governed platforms, adopted workflows and measurable business value.
Wealth ManagementEnterprise AIAgentic SystemsProduct StrategyRegulated Delivery
OptimalReasoning
Enterprise AI
Independent research
ObservePlanVerifyCorrect
Selected impact
Delivery measured in adoption.
Verified professional outcomes. Approximate figures are shown as approximate, and no financial ROI is inferred.
hours annually
Estimated productivity capacity created
European markets
Switzerland, France, UK, Belgium and Monaco
product ownership
From opportunity discovery to operational handover
A practical profile
I connect the business case, the platform and the way work actually changes.
I work at the intersection of AI, product management and enterprise delivery. I currently lead a portfolio of AI, agentic AI, workflow automation, analytics and data initiatives within wealth management at Rothschild & Co in Zürich.
My experience spans financial services, enterprise software, machine-learning products and entrepreneurship. I enjoy taking unclear problems, creating structure around them and working with technical and business teams to turn ideas into products people actually use.
More about meIndependent research
Optimal reasoning: make agents think, test and adapt.
A personal research agenda exploring how adaptive computation, external evaluation, correction memory and selective tool use could improve agent reliability and generalisation.
Explore the researchWorking hypothesis
Progress toward more general intelligence may depend on the efficiency with which a system chooses what to compute, checks its work and learns from failed attempts—not simply on producing longer answers.
01Allocate
02Explore
03Evaluate
04Correct
Quantum computing is treated as a conditional tool for suitable optimisation subproblems—not as a shortcut to AGI. Any quantum route must earn its place against a strong classical baseline.
Current mandate
AI inside regulated wealth management.
Portfolio ownership from opportunity discovery through risk-aware deployment, adoption and operational handover.
Full experienceCurrent
Rothschild & Co
AI and Automation Senior Product Manager
Zürich, SwitzerlandJune 2023–PresentIdentify and prioritise high-value opportunities, then define product direction, value cases, roadmaps, KPIs and delivery plans.
Work across COOs, divisional leaders, country use-case owners, business SMEs, IT, data teams, developers and operational stakeholders.
Build governance, release readiness, adoption planning, training, monitoring and operational handover into delivery.
Support delivery and local enablement across Switzerland, France, the UK, Belgium and Monaco.
Governance should be designed into identity, data access, permissions, evaluation and operations—not added as a final gate.
AI platform thesis
From isolated pilots to a repeatable operating system.
A practical seven-part point of view for building an AI function in wealth management.
Explore the full thesis01Strategy
02Platform
03Access
04Portfolio
05Governance
06Adoption
07Scale
“The objective is not to maximise the number of AI experiments. It is to repeatedly turn the right opportunities into safe, adopted and measurable outcomes.”
My operating principleOpportunity map
AI use cases across the private bank.
A structured catalogue spanning relationship management, investments, compliance, operations, management and technology.
Open the explorer010203040506
Relationship management
Automated meeting preparation
Relationship management
Client portfolio briefing
Relationship management
Meeting-note structuring
Relationship management
Follow-up action generation
Relationship management
Relationship-manager knowledge assistant
Relationship management
Personalised communication drafting
Research points to sustained pressure on advisor capacity, making better workflow leverage strategically relevant.[McKinsey & Company, 10 February 2025]
Selected projects
Building to understand the edges.
Independent product experiments across LLM infrastructure, voice-enabled workflows, portfolio intelligence and enterprise automation.
View all projects01LLM product infrastructure
Garlic
A contextual monetisation platform for AI applications.
An experiment in how LLM-native products can introduce useful commercial context without damaging the user experience. The product combined a Python SDK, a TypeScript interface, APIs, Supabase and vector similarity search.
- Approximately 160 users during the first four days
- Python SDK and TypeScript interface
- APIs, Supabase and vector similarity search
- Business-model exploration for LLM-native products
02Agentic workflow prototype
DocPilot
A voice-enabled AI assistant for administrative workflows.
A hands-on exploration of assistants that can understand speech, reason over a request and take bounded actions inside a workflow.
- Speech recognition and speech synthesis
- OpenAI APIs and Python
- Tool-enabled actions
- Administrative workflow automation
03Portfolio intelligence
WealthWise
A personal portfolio-intelligence and financial-analysis project.
A practical environment for combining market data, portfolio analytics and LLM-generated interpretation while keeping the distinction between analysis and decision clear.
- Python and financial data
- Portfolio analytics
- LLM-generated analysis
- Decision support, not automated advice
Field Notes
Working notes on enterprise AI.
Practical essays on platform strategy, use-case choices, governed agents and relationship-manager leverage.
Read all notesFrom AI Pilots to an AI Operating System for Wealth Management
Why private banks need a repeatable operating model connecting strategy, platform capabilities, governance, delivery and adoption.
Read field noteChoosing AI Use Cases in Private Banking
A practical framework for balancing impact, feasibility, risk, data, adoption and reuse.
Read field noteGoverned Agentic AI: Control Without Killing Speed
How identity, permissions, approvals, evaluations and observability can enable safe execution.
Read field noteA useful conversation
Building enterprise AI that survives contact with reality.
I am always interested in difficult product, platform and transformation mandates—especially where trust matters as much as speed.
Start a conversation