Problem Statement:

Currently we are required to manually scrutinize each expense ledger to determine the applicability of TDS under the Income Tax Act. This involves a time-intensive review of numerous transactions to identify whether TDS provisions are attracted, followed by a manual cross-verification with the TDS ledger to confirm whether TDS has been correctly deducted and accounted for against the respective parties and expense heads. This manual approach is prone to oversight, inconsistent application of thresholds and rates, and increased risk of non-compliance. There is a pressing need for an AI-driven solution that can automate the identification of short-deduction or non-deduction of TDS based on pattern recognition, applicable provisions, and historical data, thereby enhancing audit efficiency, accuracy, and statutory compliance.

Additionally, the risk of oversight or human error increases the chances of omissions or non-detection. If such Non/Short deduction is later identified by the A.O., clients often shift the blame onto Chartered Accountants, questioning why they were not informed at the time of audit.

Objective:

The objective of this case study is to leverage Perplexity's capabilities to automate the identification of Short/Non-Deduction of TDS from a Day-Book, ensuring compliance with various sections of the Income Tax Act. By utilizing AI-driven analysis, the goal is to streamline the verification process, reduce manual effort, and mitigate the risks of omission or human error. This approach aims to enhance accuracy, improve efficiency in TDS compliance, and provide a proactive mechanism to prevent potential disputes or liabilities arising from non deduction and payment of TDS.

AI Solution Overview:

The proposed AI solution simplifies the process of identifying TDS by leveraging AI for automated analysis. The approach involves extracting vendor-wise expenses with narration from the day-book of Tally, which is then uploaded to Perplexity for processing. Using advanced text analysis techniques, Perplexity will assess each entry against various sections of the Income Tax Act and determine whether the TDS is applicable on the expense or not.

This system integrates a synergy of cutting-edge technologies, including Natural Language Processing (NLP) for expense description analysis, Machine Learning for pattern recognition, and Knowledge Graphs to establish relational mapping between tax provisions and transaction details. These technologies work together to ensure that TDS deductions are assessed with high accuracy, detecting patterns and connections that might be missed in manual reviews.

The system will provide clear, section-based reasons for each flagged entry, ensuring transparency and compliance. The results can be structured in a Word file for easy review and documentation. This approach eliminates the need for manual verification, significantly reduces the time required for analysis, and minimizes the risk of human error.

Workflow:

Input :The process begins with extracting day-book data from TallyPrime through CA TDL by CA J Ramajayam. The key requirement is to extract the vendor wise expenses incurred with narration and vendor-wise amount of TDS deducted using Pivot Table. Following are the settings given for easy reference.

1. Processing: Once the excel file containing above tables is prepared, it is uploaded to Perplexity along with a structured prompt that instructs the AI to evaluate each entry against TDS sections of the Income Tax Act. Perplexity utilizes Natural Language Processing (NLP) to analyse expenses, Machine Learning (ML) to detect patterns in TDS deductions, and Knowledge Graphs to establish contextual relationships between tax provisions and expense categories. Based on this structured analysis, the AI systematically identifies potentially Short-Deduction / Non-Deduction of TDS and provides justifications referencing specific legal clauses.

1. Output: The AI-generated results are compiled into a structured Word file that categorizes vendor-wise expenses as TDS Deductible / Non-deductible along with a clause-based rationale. This output ensures audit-ready documentation, facilitating compliance verification and reducing the risk of tax disputes. The final report will be reviewed and be discussed with the assessee before filing TDS returns or finalizing the audit, enhancing accuracy and efficiency.

Example of Prompt used:

1. Hello. In this workbook there are 2 sheets.
2. Sheet named "TDS Deducted" is consisting the name of parties and the amount of Income tax tds deducted from them.
3. Sheet named "Expenses Incurred" is consisting of all the expense transactions of the company containing Party name, expense head, narration and the amount of expense.
4. Now your task is to check as per the Income Tax Act, on which transactions TDS is required to be deducted.
5. Then sum it for each party and compare with the data in "TDS Deducted" Sheet.
6. Do not skip any item
7. Your answer should be detailed and with explanation for each party
8. Recheck all your analysis and then only give me the final reply.

Key Metrics:

Time Efficiency: Reduction in time spent from lets say 3 Man Days for checking each and every expense and TDS applicability and verifying with TDS ledgers, to Very Few Man hours.

Improvement in structured reporting and audit preparedness and Client Satisfaction.

Challenges and Risks:

False positives or false negatives is the 1^st and foremost risk of using AI. This can be mitigated by regular fine-tuning, manual cross-validation, and hybrid rule-based checks. Another risk is that, sissing or inconsistent day-book data may affect accuracy. But it can be easily mitigated by standardising data extraction, clean descriptions, and using structured formats. AI may misinterpret complex Income-Tax provisions. While AI comes with its challenges, regular improvements and thorough validation help ensure it accurately and reliably automates TDS verification, making the process faster and more efficient.

Relevance to Chartered Accountants:

As Chartered Accountants, we are the primary go-to professionals for queries related to TDS and audit compliance. By leveraging this AI-driven solution, we can save valuable time, enhance accuracy, and complete our work more efficiently and professionally, ensuring better compliance and client satisfaction.