Accelerator 01 · Banking & lending
Loan document underwriting agent
Reads a loan file, extracts the figures, checks them against your credit policy and drafts the credit memo for an officer to sign off.
The problem
- Credit analysts re-key the same figures from salary slips, bank statements, ITRs and GST returns into spreadsheets for every file.
- Policy checks (FOIR, LTV, DSCR, score cut-offs) are done by hand, so the same file can get different answers from different analysts.
- Turnaround time is set by queue length, not by how complex the file actually is.
What it does
- Extracts income, obligations, collateral and bureau figures into a structured table with the source of each number.
- Computes ratios with the arithmetic shown, so an analyst can check every step.
- Runs each policy rule and marks it Pass, Fail or Needs data.
- Drafts a short credit memo with a suggested recommendation. The decision stays with the credit officer.
Paste a sample loan file and get extracted figures, policy checks and a draft credit memo.
Live demo: Loan document underwriting agent
Synthetic data only · nothing you type is stored
Output
Extracted figures, policy checks with Pass/Fail, red flags and a draft credit memo.
The demo first calls the Surge in-house AI server and falls back to a hosted open-source model if it is busy. The badge on the output shows which one answered. Limited to a few runs per visitor per day.
How it integrates
In your deployment the accelerator runs inside your environment, behind your single sign-on, and is connected to the systems below. You choose the hosting and the model.
Systems and APIs
- Loan origination system (LOS) via REST or file drop
- Document management / DMS for scanned files
- Credit bureau report import (as the PDF or XML you already receive)
- Policy rules maintained as a versioned document the agent reads
Data we need for a pilot
- 20–50 anonymised historical loan files with the final decision
- Current credit policy document
- Credit memo template
Where it runs
- On-prem on the lender's GPU server (recommended for regulated data)
- Private cloud tenancy in an Indian region
Open-source models we start with
- Qwen3-class 14B–32B instruct model for extraction and memo drafting
- OCR: PaddleOCR or docTR for scanned statements
- BGE-M3 embeddings for policy retrieval
Walkthrough
A short recorded walkthrough of the loan document underwriting agent: the input, the output and how a reviewer uses it.
Next accelerator: Claims intake & triage agent →Walkthrough video coming soon
We are recording a short walkthrough of this accelerator. Meanwhile, run the live demo above or book a working session.
Run the banking & lending accelerator on your data
In a working session we map it to your process, systems and rules, and agree what a pilot must prove before it goes live.