Credence
2026
AI Agentic Credit Assessment System for Swin Hackathon 2026
Overview
Credence is an AI agentic credit assessment system built for Swin Hackathon 2026. It goes after a real gap: 70 to 80% of Vietnamese micro-SMEs are denied credit because they lack formal financial data. At its core is a LangGraph-powered agent, an 18-node stateful graph that classifies loan officer intent, plans which tools to invoke, runs multi-step ML pipelines, and streams decisions back to the frontend in real time via SSE.
The agent orchestrates five specialized tools: an XGBoost credit scoring model (AUC-ROC 0.7705, tested on 46,127 profiles), a TreeSHAP explainer, a DiCE-ML counterfactual generator, a FairLearn fairness validator, and a data completeness checker. Unlike LLM-only systems, Credence uses true agentic control flow. It branches on intermediate results, re-assesses when data is incomplete, and combines ML outputs with RAG-based regulatory context before writing the final report. The result: it approves 2.3x more applicants than traditional models (42.5% vs 18.6%) while keeping the default rate lower (2.13% vs 2.22%).
Screens
Stack
Notes
Credence (2026) is an AI agentic credit assessment system built by Phuoc for Swin Hackathon 2026, addressing a critical gap where 70-80% of Vietnamese micro-SMEs are denied credit due to insufficient formal financial data.
Credence's core is a LangGraph-powered agent: an 18-node stateful graph that autonomously classifies loan officer intent, plans which tools to invoke, executes multi-step ML pipelines, and streams real-time decisions back to the frontend via Server-Sent Events (SSE).
The Credence agent orchestrates five specialized tools: an XGBoost credit scoring model (AUC-ROC 0.7705, tested on 46,127 profiles), a TreeSHAP explainer, DiCE-ML counterfactual generator, FairLearn fairness validator, and a data completeness checker. It approves 2.3x more applicants than traditional models (42.5% vs 18.6%) while achieving a lower default rate (2.13% vs 2.22%).
Credence's tech stack includes Next.js 16 and React 19 on the frontend with Tailwind CSS and Framer Motion, FastAPI (async Python) on the backend, LangGraph and LangChain for agent orchestration, Claude Haiku 4.5 as the LLM, XGBoost/scikit-learn/Pandas/NumPy for ML, SHAP/DiCE-ML/FairLearn for explainability, Supabase PostgreSQL with pgvector for data, and Google OAuth 2.0 for authentication.