How a Big Four Accounting Firm Cut Lease Agreement Review Time by 80%

With the effectiveness of new lease standards such as IFRS 16 and ASC 842, a large accounting firm faces new compliance requirements: helping clients disclose operating leases on their balance sheets. The firm's clients are mostly large enterprises, each holding thousands of lease agreements, and each agreement needs to be screened item by item to extract key information. Due to the lack of a unified format in lease agreements, the same type of information may be expressed in different terms, making keyword-based retrieval systems ineffective and posing significant obstacles to automated processing.

To address these challenges, the accounting firm integrated a solution based on "semantic folding" technology into its contract processing workflow. Semantic folding is a form of AI-based natural language processing (NLP) that encodes and represents words in binary form to capture semantic relationships between terms, rather than relying solely on literal matching.

This solution enables the firm's domain experts to quickly define and refine information extraction models—training a model requires annotating only about 50 sample lease agreements, whereas traditional methods typically require thousands of samples. After deployment, the system can extract and classify relevant information from lease agreements with a precision that manual large-scale processing cannot achieve. Meanwhile, domain experts continuously optimize the extraction process by validating, correcting, and supplementing extraction targets.

Actual Results

Ultimately, the solution reduced review and data extraction time by 80% compared to purely manual processes, significantly improved accuracy, and automatically assisted domain experts in preparing balance sheets for clients. As a result, expensive human resources were freed up to focus on more business-critical tasks. The core of the solution's success lies in its ability to clearly identify target information, classify clauses and their associated agreements, and accurately extract structured data such as lease amounts, dates, and addresses from scanned documents.

During the extraction process, the system can also use other data fragments to infer information not explicitly stated in the agreement, and import the extracted results into the client company's lease management tool. The tool then generates a summary sheet for each agreement and displays it side by side with the corresponding agreement for review, while highlighting the extracted information in the original text of the agreement, greatly improving review efficiency and traceability.

Key Selection Points

If your company is preparing to address the new lease standards, drawing on the successful experience of the aforementioned accounting firm, the following capabilities should be considered when evaluating solutions:

• Semantic-based NLP contract analysis.This type of software employs a unique, meaning-based analysis approach that reviews contracts more quickly and accurately, avoiding the limitations of keyword matching.
• Intelligent contract analysis without the need for AI experts.Some contract analysis software is designed specifically for domain experts, who can train new models for any document type starting from about 100 documents without the support of data scientists, significantly lowering the technical barrier.
• Gaining scalable insights through automated analysis.Contract analysis software should easily integrate with contract management systems and business intelligence tools, providing key information needed for risk assessment, trend identification, and discovering new revenue opportunities.

One of the most challenging aspects of the new lease standards is identifying lease clauses that are often embedded in agreements that are difficult to search. Semantic-based NLP contract analysis is an effective way to overcome this challenge quickly and accurately.