Key Project Highlights
3+
years
ongoing engagement length
99%
accuracy
Document Classification Engine
86%
accuracy
Document Entity Extraction
Client Overview
Plenti is one of Australia’s leading technology-powered lenders, offering faster, fairer loans through intelligent technology. With services across automotive, renewable energy, and personal loans, Plenti has become a trusted name in fintech, boasting 28 awards in 6 years, over 33,650 five-star reviews, and more than 15,000 customers' second loans.
Project Overview
Plenti partnered with Neurond to streamline its invoice document extraction process, enhancing operational efficiency and customer satisfaction through advanced AI automation.
Challenge
- Manual Processing Bottlenecks – Employees spent excessive time extracting and verifying data from invoices, slowing down loan approvals.
- Employee Productivity and Customer Satisfaction - A demand to improve employee efficiency and accelerate loan processing to enhance customer experience.
Solution
Neurond developed a custom AI-powered document processing system to address Plenti’s challenges, focusing on:
- Document Classification Engine: Our team trained a custom deep learning model to classify financial documents (e.g., invoices, payslips, bank statements). Unlike off-the-shelf large language models, this bespoke solution was designed specifically for Plenti’s needs, ensuring high accuracy, relevance, and security.
- Document Entity Extraction: We implemented AI-based extraction to pull key details (e.g., vehicle price, seller info, loan terms) from uploaded car invoices. This allowed Plenti to verify that invoice details matched seller information, streamlining the loan approval process.
Technologies
The Outcome
The collaboration between Plenti and Neurond delivered impressive results, transforming Plenti’s invoice extraction process, setting a new standard for efficiency and reliability in fintech lending.
- Achieved a 99% accuracy rate in categorizing financial documents, ensuring reliable and precise processing.
- Reached an overall accuracy rate of 85%, with a 95% accuracy rate for critical entities, enabling confident verification of invoice data.
- Significantly reduced manual work, boosted employee productivity and improved customer satisfaction.
