01 / Housing-loan underwriting assistance

FinCredit Copilot

Bringing documents, policy, calculations, and human confirmation into one clear lending flow.

System preview

FinCredit Copilot

A real capture of the current product interface. Select the preview to enter an available demo.

https://fincredit-copilot-demo.onrender.com/FinCredit Copilot System previewEnter system
Project overview

A lending-assistance platform for the Chinese housing-loan context, covering documents, policy, DTI / LTV calculations, risk suggestions, and human approval.

Who it is for

Housing-loan applicants, loan officers, and underwriters who need a clear view of documents, policy, calculations, and approval conditions.

What it does

  • Show role-specific application progress, documents, and conditions
  • Retrieve national policy and Chengdu rules with sources
  • Run deterministic DTI / LTV calculations while reserving the decision for a human

How to explore it

  1. Choose the borrower, loan officer, or underwriter role.
  2. Enter the corresponding demo workspace.
  3. Follow the prompts to inspect documents, run a risk profile, or ask the assistant.
  4. Leave the final credit decision to the authorized human reviewer.

Every person, amount, and approval outcome is fictional demo data and does not constitute lending advice.

Enter system

The challenge

Lending needs both natural-language guidance and deterministic calculations; a model must not replace the final credit decision.

The approach

  • Documents, Chinese statuses, approval conditions, and disclosure acknowledgements become one continuous flow.
  • National policy and Chengdu rules are managed separately, with scope and sources shown in answers.
  • Deterministic logic owns DTI / LTV calculations; the agent explains and guides.

The outcome

Delivered role-based demo interfaces for borrowers, loan officers, and underwriters, centered on policy evidence, deterministic calculations, and human confirmation.

Boundaries & reflections

All people, applications, amounts, and policy information are fictional demo data and do not represent a real lender or credit decision.