AI-powered GL Review System Improving Financial Accuracy and Speed Across Multi-Property Portfolios

About Client

A Texas-based private firm investing in and managing properties across the U.S., Canada, and Europe, focused on delivering consistent long-term value for investors. The organization operates at scale, with centralized finance oversight across geographically distributed assets and accounting systems.

Background

The client’s finance team processes large volumes of financial data every month, with a heavy dependence on General Ledger (GL) entries generated across multiple properties. These GL records are foundational to financial reporting, audits, and compliance.

Given the scale, monthly GL reviews were largely manual. While necessary, this process was slow, repetitive, and vulnerable to oversight—especially when reviewing thousands of entries with subtle contextual differences across properties and periods. An AI-powered GL approach was required to introduce intelligence without disrupting existing controls.

Challenge

Each month, thousands of GL entries flow in from systems such as Yardi. Inevitably, some entries are posted to incorrect accounts or categories due to:

  • Ambiguous or inconsistent descriptions
  • Human oversight during data entry
  • Limited visibility into historical posting patterns

Manually identifying and correcting these issues consumed significant finance team bandwidth. More critically, manual review itself could introduce fresh errors. The client needed a scalable AI-powered GL review mechanism that accelerated validation while keeping final judgment with finance.

Our Solution

The client partnered with BizAcuity to implement an LLM-driven AI-powered GL Review System designed to combine automation, explainability, and human oversight.

Secure File Upload

  • Finance teams upload monthly GL files through a Streamlit-based web application secured with Azure EntraID authentication.
  • Uploaded files are validated for structure and completeness before ingestion into Postgres, enriched with vector embeddings, and securely backed up in AWS S3 to ensure traceability and resilience.

LLM-Powered GL Analysis

  • GL entries are normalized, deduplicated, and embedded to preserve contextual meaning across descriptions and time periods.
  • Using AWS Bedrock, the AI-powered GL engine evaluates entries against historical GL data with large language models such as Cohere, Claude, and LLaMA 2.
  • The system flags potentially miscoded entries, recommends alternative accounts, assigns confidence scores, and explains its reasoning—supporting accurate, review-ready decisions.

Review and Feedback Loop

  • Finance teams receive a downloadable Excel or CSV report containing flagged entries and recommendations.
  • Users can mark suggestions as correct or incorrect and re-upload the annotated file. This feedback feeds back into the AI-powered GL pipeline, enabling continuous learning and refinement.

Decision Authority Remains with Finance

  • The system does not make changes directly in Yardi or any source system.

Final approval and posting decisions remain entirely with the client’s finance team, preserving governance, compliance, and accountability.

Outcome

The AI-powered GL review system delivered clear operational and accuracy improvements:

  • Faster monthly GL reviews with significantly reduced manual effort
  • Early identification of misclassified entries, minimizing downstream corrections
  • Cloud-native scalability across a growing multi-property portfolio
  • Continuous accuracy gains driven by real-world finance team feedback
  • Full human control, with AI supporting—not replacing—financial judgment

The new system delivers:

  • Faster monthly financial reviews with significantly reduced manual effort
  • Early detection of misclassified entries, minimizing downstream errors
  • Cloud-native design that scales seamlessly across multiple properties
  • Continuous improvement through feedback loops that make the model smarter over time
  • Finance team retains full control, with AI acting as a supportive assistant rather than a replacement

Leave a Reply

Your email address will not be published. Required fields are marked *

BizAcuity
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.