GLM 5.2 is nearly as accurate as a human book keeper (toot-books.pages.dev)

GLM 5.2 has demonstrated impressive accuracy in preparing quarterly value-added tax (VAT) returns for small UK businesses, matching human bookkeepers' performance at only a fraction of the cost—just $2.73 for processing 59 transactions in 68 minutes. This AI model executed the task using a command-line tool and accessed accounting software while adhering to specific guidelines and constraints to prevent any form of data leakage or cheating. The model's output was nearly flawless, with only a 7 pence discrepancy in the primary VAT calculation. However, it did reveal some notable minor errors, particularly in tax categorization, showcasing both strengths and limitations in the model's understanding of complex bookkeeping scenarios. This benchmarking exercise is significant for the AI/ML community as it underscores the capability of advanced AI models like GLM 5.2 to automate traditionally labor-intensive bookkeeping tasks. By successfully performing at near-human accuracy, GLM 5.2 illustrates a shift towards automating compliance-related processes for small and medium-sized enterprises (SMEs). The findings suggest that as AI technology continues to advance, it may soon be viable for startups to leverage such models extensively, redefining the landscape of financial management while lowering operational costs significantly.