Portfolio · liquidity risk · second-line review

Financial Risk Intelligence

An agentic AI system for regulatory research, liquidity risk analysis and evidence-grounded independent review.

01

Regulatory Intelligence

Search ECB, EBA and Basel supervisory material and answer with the passage the claim came from.

02

Liquidity Scenario Analysis

Run a liquidity scenario and calculate the coverage ratio in Python, including the inflow cap. The model does not do the arithmetic.

03

Independent AI Review

A second step checks that the conclusion is supported by the retrieved evidence, and refuses when the instrument is not in the corpus.

Python · LLM · RAG · LangGraph · Vector Search · Agentic AI

The briefings are analytical notes on public standards. They are not a substitute for the official text, and the desk does not give legal advice.

Try an example

Three cases from the pipeline

Interactive demonstration based on representative runs of the underlying RAG and agentic pipeline.

Choose a case. The trace on the right is the system’s execution record: retrieved passages, tool calls, and the review. It is not a model’s hidden reasoning, and this page does not call an API.

Architecture

One question, several checked steps

Hover a stage. This is the workflow the three cases replay, and the one the local pipeline implements.

Evaluation and safety

Measured on the same 60 questions

Closed book, basic retrieval, hybrid retrieval, and the full desk. Figures are the latest local run of that bench, not targets.

Basic retrieval versus the agentic desk