Latest Experiments
The first university lab dedicated to agentic finance.
City Agentic Finance Lab (CAFiL) studies how AI agents behave in financial settings through careful, reproducible experiments, from simulation to controlled pilots with partners.
Our experimental method
- Hypothesis-led studies, each with a stated question
- Agent-based and multi-agent simulation in a sandbox with synthetic customers and markets, with no live money
- Defined evaluation protocols and value benchmarks
- Controlled pilots with partners
- Human oversight and governance designed in from the start
- Reproducible methods, with results published, including what fails
- Research-ethics approval where needed
Methods across domains
Our researchers apply a range of AI and agentic methods across these fields:
- Language-model agents that use tools
- Multi-agent systems
- Reinforcement learning
- Agent-based modelling
- Retrieval-based systems
- Evaluation harnesses
The fields we study
CAFiL’s experiments focus on 20 fields.
- Retail payments and checkout Payments and commerce
- Agentic commerce and digital marketplaces Payments and commerce
- Cross-border payments and foreign exchange Payments and commerce
- Consumer lending and credit decisions Lending and banking
- SME finance and trade credit Lending and banking
- Open banking and open finance Lending and banking
- Financial wellbeing and personal finance management Lending and banking
- Wealth management and financial advice Investment and markets
- Investment research and portfolio management Investment and markets
- Capital markets operations and post-trade processing Investment and markets
- Insurance underwriting and claims Insurance and customers
- Corporate treasury, cash and liquidity management Lending and banking
- Regulatory compliance operations Regulation and assurance
- Supervisory technology and regulatory reporting Regulation and assurance
- Complaints handling and customer service Insurance and customers
- Model risk management and AI assurance Regulation and assurance
- Governance, accountability and redress for AI agents Regulation and assurance
- Agent-to-agent markets and market design Investment and markets
- Stablecoins, tokenised deposits and central bank digital currencies Payments and commerce
- Tokenised funds and real-world assets Investment and markets
All of our experiments stay within our scope statement.
Status
Our first experiments are being designed. Results will be published here.
Propose an experiment or fund a pilot
If your organisation has a question an experiment could answer, we would like to hear from you. Pilots are designed with partners and tested in a safe environment before any live use.
Or call us on +44 (0)20 7040 4963