LLM Powered Internal Audit - Evaluating Invoices

Select one vendor

Select needed columns

YOUR API TOKEN

GPT 5.2200000.2

Call LLM to analyseinvoice data andidentify split patterns

Combine all columns intoa single string

Create text blob to send to llm

Isolate Duplicate Invoices

Top 5 vendors

Isolate Approval Limit Violations

Vendor Master

Approval List

Invoices

Join Vendor data

Replace net with nothing

Replace immediate with 0

Payment terms to number

Invoice IssueLate Payment

Invoice IssueUnauthorised Vendors

Invoice Issue Duplicate invoice

Dates to datedata types

Invoice IssuesApproval Limit Violation

Combine all columns intoa single string

Join approval limits

Call LLM to analyselate paymentsBEWARE OF HAVING LLMS DO MATH. A BETTER APPROACH IS TO HANDLE ALL MATH FUNCTIONS IN KNIME, AND PASS THE DATA TO THE LLM FOR SYNTHESIS

Join Vendor Name

Format text

Get new row IDs

Create text blob

Second output shows invoices to vendors that are not in the approved list

Create text blob to send to llm

Combine all columns intoa single string

Call LLM to analyseapproval limit violations

Days to pay

Late fee

Find duplicateinvoices

Days late

Create text blob to send to llm

Late payments to first row

Vendor ID AscInvoice Date Asc

Identify instanceswhere approval limitswere breached