This is the Pythonic Accountant and today we’re going to take another look at doing some reconciliations using ChatGPT’s Advanced Data Analysis Tools. In my previous video, we took a look at reconciling two files: one with general ledger detail and one with subledger detail. However, the analysis was limited to comparing the differences and recording an entry assuming the subledger was correct.
To make the analysis more interesting, we decided to add additional columns to the files. This allows us to explore root causes using generative AI for root cause analysis. We uploaded the two new files, SU Ledger updated and general ledger updated, and requested a reconciliation. If discrepancies are found, we want the AI to perform root cause analysis and recommend corrective actions, both short-term in a journal entry and long-term in systemic fixes or trainings.
The AI analyzed the data and identified discrepancies between the subledger and the general ledger for multiple dates. However, it failed to provide any actual root cause analysis in the issue paper. We requested the AI to look at the columns we provided and pick up on the discrepancies. It then revised its findings and identified issues such as returns and cut-off issues. These findings may indicate specific issues affecting the financial data.
We then requested the AI to provide the actual dollar amounts for each root cause in a table, along with the journal entry needed to correct the discrepancies. The AI recommended a credit to sales revenue and a debit to adjustment. Although it didn’t specify the debit account, it provided a starting point for further analysis.
Overall, the AI’s analysis was not perfect, but it was not terrible either. The data provided to the AI was not very detailed, with only one column indicating the root cause. However, this demonstration should inspire you to explore using Advanced Data Analysis Tools for reconciliations. If you found this video helpful, please like and subscribe for more content.
