AI will not replace the Chartered Accountant. It will replace the CA who sells time cost instead of Value based judgement. For a profession of over five lakh members, that distinction now decides growth of any Indian Chartered Accountant.

The signals are no longer speculative. At its AI Innovation Summit (AIS) in June 2026, ICAI said it had trained more than 50,000 members in AI and built over 150 GPT-based tools, signed an MoU with Sarvam AI, and launched an advanced AI certification (AICA Level 3). ERP vendors, meanwhile, are embedding agents that close books, reconcile accounts and prepare tax filings.

The opportunity is the mirror image of the threat. Work that took a team around a week, now takes one CA a day or even a couple of hours . A small or mid-sized practice can therefore take on mandates that were once the preserve of Big Four firms and boutique banks.

1. Traditional practice: do the same work for more clients, and sell new work to each

In tax, audit, accounting and shared services, AI cuts the effort per engagement by half or more. The smart CA does not pass the entire saving on as lower fees. The freed capacity goes into serving more clients and adding advisory layers that were never economical before

2. High-end advisory: AI lowers the entry barrier

Investment banking, M&A, valuation, insolvency and arbitration were once closed to small practices because they needed large analyst teams. AI is that analyst team. A two-partner firm can now produce the first cut of work that used to need eight people and compete on judgement and relationships.

• Investment banking and fund-raising. AI drafts information memoranda and teasers from management data, builds buyer and investor long-lists from public filings, and screens comparable deals. The CA's edge is the equity story, the negotiation and the network.

• M&A and due diligence. AI reads thousands of data-room documents in hours, flags change-of-control clauses, related-party flows, contingent liabilities and GST or TDS exposures. The CA decides what is material, quantifies it and translates it into price and Share Purchase Agreement protections.

• Valuation. AI assembles comparables, normalizes financials, runs DCF sensitivities and drafts the report. Registered Valuers remain answerable for assumptions, so the value lies in defending the discount rate, the terminal growth and the synergies before a tribunal, regulator or buyer.

• Insolvency and restructuring. For Insolvency Professionals, AI collates and verifies claims, reconciles creditor lists, runs forensic tests for preferential, undervalued and fraudulent transactions under the IBC, and drafts CoC agendas. Faster process work means more mandates handled within the IBC timelines.

• Arbitration and expert witness work. AI organizes the record, builds chronologies, and models damages under multiple scenarios. The CA as quantum expert or arbitrator brings independence and credibility, which no model can supply.

The pattern is the same in each area: AI does the gathering and first drafting; the CA owns the opinion, the number and the accountability. That accountability is what clients pay a premium for.

3. Research: From searching to Asking the right questions to AI tools.

Research is where AI changes the CA’s Day most visibly. The old workflow was search, read, write and draft. The new one is ask, verify and decide.

Three disciplines make AI research safe to rely on:

1. Verify every citation against the primary source. Models can invent case names, section numbers and circulars with full confidence. An opinion resting on a non-existent judgment is a professional misconduct risk.

2. Prefer grounded tools. Assistants that answer from a defined library (ICAI's CA GPT and its GPT tools, licensed tax databases, the client's own documents) are far more accurate than open chatbots.

3. Keep a research trail. Save the prompts, sources and the CA's own conclusions. This becomes audit evidence and protects the CA if the opinion is challenged.

The net effect is that research stops being a cost center billed by the hour. It becomes the raw material for faster, better-reasoned opinions, and the CA who asks sharper questions is the CA who wins.

4. ERP, point solutions and platforms: AI blends in, it does not replace

AI will not eliminate ERP, and it will not eliminate every specialist tool. It will change which tools survive. The ERP remains the system of record, the books that auditors, tax officers and lenders rely on. AI becomes the layer that reads, reasons over and acts on that record.

The ERP vendors are already moving this way. At Sapphire 2026, SAP announced 224 AI agents and 51 domain-specific Joule assistants, with finance getting the deepest coverage across closing, planning, billing, receivables and tax compliance . SAP estimates one of its reconciliation agents can save up to 70% of manual reconciliation time . Oracle has embedded AI agents across Fusion Cloud finance, HR and supply chain. Equally important, SAP's agents can now talk to agents in third-party systems through open protocols such as MCP (Model context protocol) and A2A (Agent to Agent).

What this mean for the tool landscape:

• Thin point solutions are at risk. Tools that only extract invoice data, prepare a standard reconciliation or format a report can be replicated by a general AI model or an ERP agent.

• Deep point solutions survive. Tools built on proprietary data, regulatory integration or certification (GST and e-invoicing GSP connectivity, licensed case-law databases, audit software with methodology built in, valuation data providers) stay valuable. AI makes them more useful, not obsolete.

• Platforms become orchestrators. The winning setup is ERP as the record, specialist tools for regulated or data-rich tasks, and an AI layer that connects them and handles the work in between.

• Data quality decides everything. Agents fail on fragmented, inconsistent master data. Cleaning and governing that data is itself a large advisory opportunity for CAs.

For the CA, the role shifts from tool user to architect and controller. Clients need someone who can choose the stack, design controls over AI agents that post entries and file returns and certify that the outputs are reliable. That is a natural extension of audit and internal-control expertise.

5. Will AI eliminate CA jobs in FP&A, tax computation and Excel work?

It will eliminate tasks, not the profession. But the honest answer is that a large industry built on computation, reconciliation and spreadsheet assembly will shrink sharply, and junior roles will feel it first.

The real risk is to the training pipeline. Articled assistants and young CAs learned the craft through computation and reconciliation. If machines do that work, firms must deliberately teach judgement through review of AI outputs, case-based training and earlier client exposure. ICAI's move to embed AI and data analytics across the CA curriculum through its Committee for Review of Education and Training is the right direction.

The CA who was valued for building the spreadsheet must now be valued for what the spreadsheet means. FP&A professionals who move from preparing numbers to interpreting them will be more in demand, not less.

6. Laws that need updating to protect the CA's work

India has chosen not to write a standalone AI law for now. Ministry of Electronics and Information Technology’s AI Governance Guidelines of November 2025 rely on existing laws and sector regulators. That places the burden on the CA Act, the IT Act and the Copyright Act to deal with AI-enabled impersonation, plagiarism and misuse. Its better if a standalone law on AI emerges in next couple of years.

Where things stand. The Income Tax Rules amended in Budget 2026 now require labelling and fast takedown of deepfakes and other synthetic audio-visual content, but AI-generated text is outside that definition. A forged AI-written audit report or tax opinion is therefore not covered by the new labelling regime. On copyright, DPIIT's December 2025 working paper proposes a mandatory blanket license for AI training on lawfully accessed works, with royalties collected centrally; a second paper on AI outputs is awaited.

The suggested changes are not pending legislation. ICAI does not need to wait for Parliament on most of them: practice standards, ethics guidance and UDIN upgrades are within its own powers.

7. A practical playbook for CAs

1. Reprice from hours to outcomes. Fixed fees, retainers and success fees let the CA keep the productivity gain instead of giving it away.

2. Pick one high-end niche. Valuation, insolvency, transaction support or dispute quantum. AI gives a small firm bandwidth; specialization gives it credibility.

3. Build a governed AI stack. Enterprise-grade AI with data controls, the client's ERP, and specialist tools for tax, audit and data. Write a firm-wide AI usage policy.

4. Make review the core skill. Every AI output is a draft. Train teams to check citations, numbers and logic, and record that review.

5. Selling AI assurance. Controls over ERP agents, model validation and AI governance audits are a new attestation market built on skills CAs already have.

6. Protect your own work. Digitally sign reports, watermark opinions, register firm brands, and keep confidential material out of public AI tools.

7. Invest in people differently. Use ICAI's AI certification programmes and teach juniors judgement early, since computation no longer trains them.

Conclusion

Every wave of technology, from computerized accounting to GST and faceless assessment, has been predicted to shrink the profession. Each time, CAs who adapted grew. AI is a bigger wave, because it takes over not just calculation but drafting and research. Yet it cannot take on accountability. Clients, courts, regulators and investors still need a designated and certified professional, who stands behind the number. The CA who uses AI to do more, faster, and then signs with confidence will not lose relevance. That CA will expand into rooms once reserved for investment banks and global firms

More stories

What the Travel Industry Taught Me About the Future of the CFO

Shareholders Letters & Wisdom from Warren Buffett

Finance teams in GCCs need to stop waiting for instructions