1st Place
A self-supervised deep learning pipeline that treats loan payment history as a sequence – like an LLM for credit – to predict defaults up to 6 months earlier than traditional scoring.
2nd Place
An SF-native agent framework that compiles prospectuses into typed, executable deal models – letting deterministic primitives do all the maths so the AI only orchestrates, never hallucinates.
3rd place
RetroFin AI guides Spanish homeowner communities through retrofit financing end to end – from uploading building accounts to generating a signed loan application – using live Euribor data, subsidy stacking, and AI-generated legal documents.
An open, structured-finance agent framework. Clean contracts, a dynamic planner that composes audited primitives into an inspectable JSON DAG, and a verifier that refuses to ship an answer whose citations don’t resolve to a real source page or row.
A runtime executing declarative workflows to execute primitives vs a prospectus doc.
An Agentic AI system that provides proactive climate risk insights for mortgage portfolios, enabling banks to identify vulnerabilities before climate events occur.
An AI-Powered Disaster Risk Dashboard anticipating systemic risks (e.g., climate disasters) that destabilize loans, insurance, and investments.
An invoice financing platform with an AI agent handling credit scoring. Includes alternative data ingestion from web search.
Issuria is a web application that enables bankers to rapidly generate PRIIP- compliant Key Information Documents (KIDs) for structured financial products using natural language.
TWBRIYPR is a human-readable config layer for payment routing logic that enables payment specialists to define fallback rules in a shared Google Doc with no code.
.png?width=567&height=140&name=086ab-algoritmica-logo%20(1).png)