Operating Model Simulator
Size and stress-test an operating model before you test it on people. Discrete-event engine, Monte-Carlo runs, P10-P50-P90 range bands on every answer - not a single guess.
aibanklab.com
One shelf for everything I build in public on AI transformation for banking - interactive simulators, governance tooling and regulator conformity mapping. Every artifact runs in your browser and sends nothing anywhere. Every example is synthetic and fictional by design.
The shelf grows over time - new artifacts land here first.
Working tools you can use today. No sign-up, no data leaves your browser.
Size and stress-test an operating model before you test it on people. Discrete-event engine, Monte-Carlo runs, P10-P50-P90 range bands on every answer - not a single guess.
Price a process three ways - human as-is, human optimized, AI with a human in the loop - including a hallucination tax for the cost of AI being wrong. Break-even, payback and where to draw the line.
The current wave - agentic governance made tangible.
Configure a fictional banking AI agent - spend ceiling, delegation depth, tool scope - then run a simulated day and watch the controls fire, the audit log build and the conformity scorecard fill in.
An interactive map from published AI guidance to concrete agent controls - filter by jurisdiction, pillar and binding status, with every cell linked to the exact public source paragraph.
An operator standard for agentic AI in regulated finance: five pillars expanded into leader-checkable controls with stable IDs, plus an honest five-jurisdiction conformity mapping.
Next on the shelf.
A weekly research radar on AI in banking - implementations, policy moves and papers worth reading - published here as a browsable archive.
The crosswalk, written up as a citable practitioner paper with method and sources.
Personal work, personal views. All scenarios, numbers and examples across the Lab are synthetic and fictional. Sources referenced are public documents only.