I recently filed a fairly complicated income-tax return myself with the assistance of Claude. A practising Chartered Accountant reviewed it afterwards and identified certain issues, improved the presentation and gave me the confidence of an independent professional review.

But that was not my principal takeaway. What struck me was how significantly AI had narrowed the gap between what I could understand and prepare myself and what a trained professional could add. That gap is narrowing rapidly.

Artificial intelligence entering the profession is no longer a prediction — it has already happened. The real question is which functions will be automated first and which will remain dependent on human judgement.

Transaction advisory: A democratised commodity

Transaction advisory is one of the first areas where AI is already creating a serious challenge. Today, a taxpayer may pay anywhere from ₹25,000 to several lakhs for advice on a complicated GST, income-tax or commercial transaction. The fee is justified because the professional is expected to understand the facts, research the law, identify risks and provide a defensible position.

But specialised AI applications are already performing this function. I have personally seen applications such as Vidur, Taxmann AI deal with complicated transactions and produce answers that were perfect. The answer may still require professional review, but the amount of independent thought required from the professional is reducing.

The concern that a normal AI model may hallucinate is valid. But it does not apply equally to every specialised application. An AI system trained on a controlled and well-indexed legal database, with citations, cross-referencing and human training, can become more reliable than an ordinary professional working under time pressure.

Transaction advisory is therefore becoming a democratised commodity. It will not disappear, but it will become available at a fraction of its present cost.

AI-accountants will take over accounting

For decades, Chartered Accountants have built accounting systems, designed ledgers and prepared balance sheets and profit-and-loss accounts. This work remains important. But a significant part of accounting is structured, repetitive and rule based.

A business raises an invoice. Purchases are booked. Payments and receipts are recorded. Expenses are classified. Reconciliations are performed. Ledgers are maintained. Financial statements are generated.

In a business with a limited number of ledgers and a predictable transaction flow, this entire process can be handled by an AI-enabled accounting agent. The agent can read invoices, classify transactions, identify missing documents, reconcile bank entries and prepare the financial statements. Once such systems become affordable and easy to use, routine accounting work will move rapidly from professionals to agents.

Integrated automated compliance

The next step after accounting is compliance. The core compliance work of most businesses is concentrated under three major laws: the Companies Act, the Income-tax Act and the GST law.

In each case, the business is required to maintain data in a prescribed manner and submit that data to the Government through a prescribed form. At present, a substantial professional fee is charged for converting this information into returns, statements and statutory filings. This is, in many cases, sophisticated Excel work. It requires knowledge of the law, but it also involves recurring data maintenance, classification, reconciliation and submission.

An AI agent that handles the entire flow from invoicing to purchase booking can also maintain the accounts and complete the related compliance. The agent will know the transaction history. It will understand the tax treatment. It will identify mismatches before filing. It will prepare reconciliations and explain why a particular figure has been reported.

The present model often requires the taxpayer to maintain one system for invoices, another for accounts, a third for GST and a professional office for filing. The future model will integrate all of these.

Legal research will be AI-driven, but litigation will remain human-dependent

Legal research is already being transformed by specialised AI applications. Platforms like CaseMine and VetoAI are applying AI to tribunal and court databases to produce summarised case law insights and argument-ready research notes. A junior professional who previously spent two days doing legal research may be able to obtain a well-organised research note in minutes. The senior professional will then spend time reviewing the argument, testing its weaknesses and developing the litigation strategy.

However, advocacy before courts and tribunals will remain a physical and human function for much longer because it involves real-time discretion in interpreting facts, engagement with the bench and the ability to respond instantly to unexpected questions or judicial concerns. The manner of presentation, tone, persuasion and reading of the courtroom are critical to outcomes and these depend on human presence and judgement.

There is already a visible shift among Chartered Accountants towards law and litigation. Many are pursuing law degrees and, in some cases, surrendering their certificates of practice to enter legal practice formally. This is partly an economic response: as AI reduces the value of routine compliance and advisory work, professionals will naturally move towards higher-value functions such as advocacy, strategy and dispute resolution.

The B2C opportunity no longer exists

Income-tax filing has already become a commodity for a large part of the population. Online platforms offer return preparation at very low prices. Businesses such as ClearTax have already changed the way individuals and small businesses file their returns.

The continued involvement of a physical CA is often based on trust, reassurance and the fear that an AI system may make a mistake. That trust is important. But trust is not a permanent defence against technology.

As Government departments open their systems and APIs for secure access, an AI agent will be able to obtain the relevant data, prepare the return, identify inconsistencies, calculate advance tax, make the payment and complete the filing. It can also retain the records and remind the taxpayer of future obligations.

As Gen Z becomes the dominant generation of taxpayers, trust will gradually shift from the “trusted CA” to the “trusted AI agent”. For nearly 90% of routine, uncomplicated returns, that agent will be able to independently obtain the relevant information, prepare the computation and file the return. Human review will increasingly be reserved for complex facts, uncertain legal positions and high-value cases.

What will remain for the professional?

The answer is not nothing. Professionals will continue to matter where facts are contested, the law is unsettled, the financial exposure is significant, or the outcome depends upon advocacy and persuasion. They will also remain important for governance, ethical judgement and statutory signoffs.

But occasional hallucinations cannot be the profession’s defence against AI. The answer lies in reliable databases, source verification and professional oversight. Chartered Accountants understand law, finance, systems and business and have consistently adapted to technological change. Ironically, their adaptability may accelerate the disruption of their own profession.

The AI-CA is not a distant possibility waiting to be invented. Its individual functions already exist. They simply have not yet been integrated into one trusted agent.

 

—The author, Siddharth Jain, is a former IRS officer. The views expressed are personal.