Artificial intelligence investments must be evaluated against tangible business outcomes, but CFOs also need to recognise that some strategic technology initiatives may not generate immediately measurable returns, finance leaders said at the inaugural ETCFO Annual Conclave 2026.
Speaking during a panel discussion titled “The AI Cost Equation: Balancing Investment, Efficiency and ROI,” senior executives from JSW Steel, Dr. Reddy’s Laboratories, Mankind Pharma, ClearTax and Ripple Treasury examined how organisations are moving beyond AI experimentation towards enterprise-wide adoption while maintaining financial discipline.
The discussion, moderated by Saurabh Kumar, Consulting Editor, ETB2B, brought together Swayam Saurabh, CFO, JSW Steel; M.V. Narasimham, CFO, Dr. Reddy’s Laboratories; Ashutosh Dhawan, Global CFO, Mankind Pharma; Sumit Kawariya, Head Finance & Transformation, ClearTax; and Rob McGuinness, Managing Director, APAC & ME Sales, Ripple Treasury.
The panel examined a central challenge confronting finance leaders: how to distinguish between AI investments that deliver immediate productivity improvements and those that build capabilities essential for future competitiveness.
While participants acknowledged that AI adoption is accelerating, they emphasised that its financial impact varies considerably across business functions, industries and implementation stages.
JSW Steel prioritises AI investments based on business impact
Swayam Saurabh outlined how JSW Steel is categorising AI investments according to their potential to improve operational efficiency and deliver measurable financial benefits.
He identified back-office processes as an immediate opportunity, including transaction processing, GST reconciliation, fraud prevention and internal auditing.
AI-enabled systems, he explained, can improve efficiency while enabling more comprehensive transaction analysis and audit coverage.
However, the larger opportunities for JSW Steel lie in complex industrial operations.
According to Saurabh, logistics accounts for approximately 18% of the company’s costs, while its operations involve moving around 150 million tonnes of material annually through road, rail, waterways and slurry pipelines.
Optimising transportation routes and logistics networks therefore represents a significant opportunity for AI-driven cost reduction. Another potential application involves improving the blending of raw materials used in blast furnaces.
Different grades of iron ore contain varying mineral compositions, making it necessary to determine combinations that optimise furnace output, fuel consumption and operating costs. Saurabh said the company is pursuing both relatively straightforward applications that can generate near-term savings and more ambitious projects requiring longer development periods.
On investment approvals, he distinguished between foundational technology requirements and discretionary AI initiatives. Data quality, compliance and establishing a reliable source of information are essential capabilities that may not always require conventional financial justification. Other AI investments, however, should be evaluated using established capital expenditure measures, including internal rate of return and payback periods.
For large and complex projects, he advocated milestone-based approvals to assess whether expected outcomes are being achieved before committing additional capital. This approach enables companies to pursue ambitious transformation initiatives while containing financial exposure.
Dr Reddy’s demonstrates the impact of finance automation
M.V. Narasimham, CFO, Dr. Reddy’s Laboratories, presented an example of how technology-led transformation can improve financial reporting and decision-making. He explained that the company previously required approximately six to seven days to close its monthly financial books, creating delays in financial reporting across its geographically diverse operations.
With operations spanning more than 80 countries, the company faced significant complexity in consolidating financial information. Narasimham said Dr. Reddy’s initiated a major automation programme in October 2023, working with PwC to redesign and automate financial processes. Following approximately 16 months of implementation, the company achieved a substantial improvement in its monthly closing cycle.
“From March 2025, we are closing books on day two by 10 a.m. or 11 a.m.,” he said.
Importantly, this improvement was achieved without stopping routine business transactions, including billing and manufacturing operations. The transformation also enabled the company to automate approximately 98–99% of its financial dashboards, reducing dependence on manual reporting.
Finance and business teams can now access information on sales performance, margins, operating expenses and profitability across business units and manufacturing facilities. Narasimham emphasised that the benefits extend beyond finance, enabling more timely business reviews and better operational visibility.
He also highlighted the importance of data governance, particularly maintaining accurate master data, in ensuring that automation produces reliable results. While discussing return on investment, Narasimham acknowledged that some technology programmes require management conviction before their financial benefits can be fully quantified.
He argued that organisations must sometimes invest in foundational transformation initiatives without expecting complete ROI visibility at the outset. However, he stressed that subsequent investments should increasingly demonstrate measurable improvements in operating costs, productivity and profitability. The company is also deploying AI across research and development, manufacturing quality, supply chains and marketing.
Narasimham cited improvements in quality investigations, where technology-enabled processes have reduced the time required to investigate certain product specification issues from weeks to days.
Mankind Pharma adopts a three-part AI investment strategy
Ashutosh Dhawan, Global CFO, Mankind Pharma, described how the company has organised AI adoption around three strategic priorities: strengthening its core business, transforming support functions and building capabilities for the future. The first priority focuses on sales and marketing, where AI is being used to improve the effectiveness of medical representatives and tailor communications to healthcare professionals.
Dhawan explained that increasingly limited attention spans have made concise, targeted product communication important. AI can help sales teams identify appropriate products and therapy-related messages for different healthcare professionals while improving the effectiveness of short-format presentations.
The second priority involves transforming functions such as finance, human resources, information technology and supply chain management. These applications include forecasting, working capital optimisation and operational efficiency improvements.
The third area focuses on long-term product development, where pharmaceutical companies face extended research cycles and complex regulatory and intellectual property considerations. Dhawan said Mankind Pharma has partnered with a US-based AI-native company to explore ways of accelerating formulation development and identifying potential patent-related issues earlier in the process. Such applications can reduce the need for repeated development cycles, although their financial returns may take longer to materialise.
He also emphasised that successful adoption requires employees to become comfortable using AI tools. Mankind Pharma has focused on providing employees, particularly at junior and middle-management levels, with access to AI tools before introducing more extensive process changes. The objective is to reduce resistance, build familiarity and encourage employees to participate in identifying practical applications.
Dhawan highlighted the use of agentic AI in the company’s tender business, including matching tender requirements with available company information and improving the identification of relevant opportunities.
These applications, he said, provide clearer opportunities to measure operational improvements and business growth than longer-term research initiatives.
ClearTax shifts finance reporting from information to decision-making
Sumit Kawariya, Head Finance & Transformation, ClearTax, argued that the real value of AI lies in moving beyond static reporting towards actionable business intelligence.
He explained that traditional management information systems often provide historical numbers and dashboards without clearly identifying the decisions organisations need to make. ClearTax has therefore adopted a decision-oriented approach to financial reporting.
Rather than simply displaying financial information, AI-enabled systems can identify emerging business risks, operational bottlenecks and areas requiring management attention.
Kawariya described this as a shift from a management information system to a decision-making system. AI can analyse large volumes of financial and operational information continuously, helping finance leaders prioritise issues that may otherwise remain hidden within routine reports. He also shared an example of how AI-enabled services have improved access to income tax filing for gig workers.
According to Kawariya, ClearTax introduced a WhatsApp-based tax filing solution to address the difficulty of reaching workers who rarely use conventional tax filing websites.
He said more than one million gig workers filed income tax returns through WhatsApp during the previous year, with approximately 70% filing returns for the first time. The initiative also helped process approximately Rs 70 crore in tax refunds for users, he said. The example demonstrated how AI can create value beyond internal cost savings by simplifying access to financial services and reaching previously underserved groups.
On enterprise adoption, Kawariya observed that companies are increasingly creating innovation budgets to accommodate AI experimentation. Rather than committing large amounts of capital immediately, organisations are testing use cases through shorter implementation cycles before scaling successful applications. He cited initial evaluation periods of approximately 90 days, followed by longer-term assessments of whether the technology delivers intended outcomes.
Kawariya also argued that companies must establish clear ownership of AI-generated benefits and incorporate expected efficiency gains into future operating budgets. Without accountability and measurable financial targets, he warned, organisations may struggle to translate technological improvements into actual returns.
Treasury AI is moving towards agentic workflows
Rob McGuinness, Managing Director, APAC & ME Sales, Ripple Treasury, discussed how AI adoption in corporate treasury has evolved from early machine-learning applications towards more sophisticated generative and agentic solutions.
He recalled that earlier attempts to use machine learning for cash flow forecasting often relied heavily on historical bank account information, limiting their ability to capture the full complexity of business operations.
More recent AI applications, he said, have improved the analysis of large datasets and strengthened forecasting capabilities across cash management, foreign exchange exposure and interest rate risk. The growing use of agentic AI, he said, makes transparency and traceability increasingly important.
Businesses must be able to understand how automated decisions are made and determine where AI-generated recommendations should be subject to additional controls.
Measuring AI ROI requires different approaches
A recurring theme across the discussion was that conventional return-on-investment calculations may not adequately capture the value of every AI initiative. While transaction automation, forecasting improvements and operational cost reductions can produce relatively clear financial benefits, investments in research, data infrastructure and employee capabilities may take longer to generate measurable results.
The panellists suggested that CFOs should distinguish between foundational capabilities necessary for enterprise readiness and projects designed to generate specific financial outcomes. For investments with uncertain returns, phased implementation and milestone-based funding can help balance experimentation with financial discipline.
Narasimham argued that the absence of immediate ROI visibility should not automatically prevent strategic investments, particularly when they can materially change how an organisation operates. Dhawan similarly noted that pharmaceutical research applications may involve several interconnected development stages, making it difficult to isolate the financial contribution of AI.
Kawariya emphasised the importance of considering indirect benefits, including reduced management time, faster approvals and improved decision-making. He noted that traditional ROI calculations may underestimate the value of automation when they focus only on direct processing costs while overlooking time spent coordinating and following up on transactions.
Predictability, governance and human oversight remain priorities
When asked what additional capabilities CFOs would most value from AI, the discussion highlighted predictability as a significant priority.
Saurabh pointed to the need for better visibility into quarterly performance, cash flows, cost movements and potential financial surprises. Narasimham highlighted the opportunity for AI to analyse a wider range of operational and financial data rather than relying primarily on selected indicators. Dhawan emphasised that beyond generating insights, AI should help decision-makers understand the potential consequences of different actions.
