Loan Application Underwriting (Studio RPA, Coded Agent, Coded App, Maestro)¶
Use the UiPath coding skills to build a complete code-first underwriting pipeline for UiBank
Overview¶
UiBank receives loan applications every day. Each one needs a credit assessment — income check, spending analysis, debt-to-income ratio — before a loan officer can make a decision. Doing this manually takes time, introduces inconsistency, and doesn't scale.
This exercise takes a code-first approach to the problem. Instead of configuring low-code workflows, you describe what you want to build in plain English — and your coding agent, equipped with UiPath skills, generates each component for you. You'll see exactly when code beats drag-and-drop, and why.
You'll build four components in this order: the coded agent first, then the loan officer app, then the Maestro flow that wires them together, and finally the RPA workflow that drives the whole process. Each lesson is self-contained — by the end, they all run together.
The use case: a CSV of loan applicants is processed row by row. For each applicant, the RPA submits a loan application to the UiBank API and immediately starts a Maestro process with the returned loan ID. Maestro runs the AI credit assessment, routes low-risk applications to an automatic approval, and sends medium and high-risk ones to a loan officer for review.
What You'll Build¶
Four components, built in this order:
- Coded Agent — a Python LangGraph agent that receives a
loan_id, calls the UiBank API to verify the quote and retrieve market data, computes financial metrics, and uses the LLM to produce a credit score, risk band, and assessment summary - Loan Officer App — a React Coded Action App that presents the AI assessment to a loan officer and captures an Approve or Reject decision
- Maestro Flow — a UiPath Flow process that calls the agent, auto-approves low-risk results, and opens the loan officer app for medium and high-risk cases
- RPA Workflow — a Studio XAML workflow that reads
loan-applications.csv, submits each row to the UiBank API, and starts the Maestro Flow process for each accepted application
POST /api/quotes/newquote → UiBank APIloan_idrisk_band · credit_score · assessment_summaryresults.csv → upload to LoanUnderwriting storage bucket
Before you start¶
| What's a Coding Agent? | Understand the mental model, tools, skills, and how UiPath fits in |
| Installation Guides | Install Node.js, Python, .NET SDK, UiPath Studio, Git, and a coding agent |
| Install UiPath Skills | Install the UiPath CLI, authenticate, and load the skills into your coding agent |
Steps¶
| Step | Focus |
|---|---|
| 1. AI Underwriting Assessment | Use the uipath-agents skill to build a Python LangGraph agent that analyses applications and produces a structured credit assessment |
| 2. Loan Officer Review Dashboard | Use the uipath-coded-apps skill to build a React Coded Action App for loan officers to review AI assessments and approve or reject applications |
| 3. Orchestrating with Maestro Flow | Use the uipath-maestro-flow skill to wire the coded agent and loan officer dashboard into a Maestro process with conditional routing by risk band |
| 4. Application Intake & Enrichment | Use the uipath-rpa skill to build a Studio XAML workflow that submits applications to UiBank and writes results to a CSV |
What you need¶
- A coding agent installed (Claude Code, Codex CLI, Gemini CLI, etc.)
uipCLI installed (uip --versionshould return 1.195.1 or later)- Node.js LTS (for the UiPath CLI and the React app)
- Python 3.10+ (for the coded agent)
- UiPath Studio installed locally (to open and run the generated XAML)
Training Environment
Log in at https://cloud.uipath.com/tpenlabs and use tenant CodingAgentsPractice for this exercise.