Research & evidence
External claims on this website are grounded in original publications from regulators, international institutions and established industry research organisations.
Links checked 30 July 2026I use research to challenge and sharpen my judgement—not to replace it. Dates below refer to the original publication or the current official update where stated.
22 primary or institutional referencesAdvisor capacity and productivity
Connects rising demand for advice with a need to improve advisor leverage through technology and operating-model change.
Original sourceLong-term wealth-management change
Frames AI, demographic change and client trust as connected forces shaping the next decade.
Original sourceIndustry structure and advisor technology
Explains why technology, workflow redesign and change management are becoming core capabilities for wealth managers.
Original sourceAI strategy and scaled execution
Argues that banks should anchor AI in business strategy, prioritise value and move beyond disconnected pilots.
Original sourceGlobal wealth-management growth
Provides current context on changing growth dynamics and the need for more productive, differentiated wealth propositions.
Original sourceAI value realisation
Emphasises impact-led use cases, implementation discipline and sequential scaling rather than proof-of-concept volume.
Original sourceSwiss financial-sector AI adoption
Provides direct evidence on adoption, strategy, generative AI use and third-party dependency across supervised Swiss institutions.
Original sourceSupervisory expectations
Sets out a risk-based view of governance, responsibility, data, model, third-party, legal and reputational risks.
Original sourceFinancial-sector AI regulation
Highlights governance, skills, model risk, data governance and third-party providers as practical areas needing attention.
Original sourceAI governance and implementation
Offers a practical governance and risk-management framework for institutions handling critical functions and sensitive data.
Original sourceGenAI in Swiss banking
Translates the opportunities, regulatory context and enabling conditions for generative AI into the Swiss banking setting.
Original sourceEuropean AI regulation
Provides the official risk-based framework and current application timeline relevant to cross-border European operations.
Original sourceResponsible AI transformation
Examines how financial institutions can connect value creation with governance, workforce and operating-model change.
Original sourceIntelligence as skill-acquisition efficiency
Provides the conceptual basis for measuring generalisation, priors and the efficiency with which a system acquires new skills—not only its score on familiar tasks.
Original sourceAbstract reasoning and efficient generalisation
Introduces a benchmark designed to remain relatively easy for humans while exposing capability and efficiency gaps in frontier reasoning systems.
Original sourceAdaptive inference-time computation
Shows that the value of additional reasoning compute depends on the problem and the strategy used, supporting selective rather than uniform allocation.
Original sourceLimits of intrinsic self-correction
Finds that asking a model to reconsider its own reasoning without external feedback can fail or degrade performance, motivating independent evaluation signals.
Original sourceFeedback, memory and agent correction
Demonstrates an agent design that converts task feedback into linguistic reflections and episodic memory to improve later attempts across several task types.
Original sourceStructured access to tools and resources
Defines capability negotiation and the resources, prompts and tools that an MCP server can expose. It is an integration layer, not a reasoning or safety mechanism.
Original sourceQuantum combinatorial optimisation
Introduces QAOA as a method for producing approximate solutions to certain combinatorial optimisation problems—the narrow technical basis for exploring quantum-assisted planning subproblems.
Original sourceQuantum advantage and benchmarking
Reviews the opportunity and open questions in quantum optimisation, emphasising rigorous comparison with appropriate classical methods before claiming advantage.
Original sourceQuantum constraint solving and current limits
Examines QAOA on hard constraint-satisfaction problems while documenting the substantial practical and comparative hurdles involved in outperforming classical solvers.
Original source