Ankit Gupta, Managing Director, Protiviti Member Firm for India highlighted that while AI-ERP adoption is gaining momentum across industries, success depends less on technology deployment and more on organizational readiness, workflow integration, and governance.
Q1. Your survey covers companies across many industries in India. In which sector do you think AI-ERP adoption is moving the fastest, and which sector is most at risk of being left behind?
Ans: Our survey shows that momentum is strongest in manufacturing, automotive and consumer goods. Many respondents in these sectors report moving beyond pilots to scaled deployments, particularly in demand forecasting, production planning and supply chain optimization. The case for AI in ERP is highly compelling given the high operational complexity and clear impact on margins that are driving this progress. By contrast, public sector enterprises, utilities and health are moving more cautiously. Many respondents in these areas are still in the early exploration or are in the process of stabilizing ERP and consolidating data. The usual problems are existing legacy systems, data silos, and regulations. According to the survey, the major difference lies not in awareness but in readiness. While confidence in AI's capabilities is high in all sectors, companies with significant cloud ERP systems, standard procedures, and quality data can implement their intentions faster. To conclude, the implementation of AI-ERP is rapid when the digital environment is well-established, whereas it is slow in cases of nascent data and procedural maturity
Q2. The report highlights the gap between what leadership intends and what employees actually adopt on the ground as the biggest risk. What is the one thing companies most commonly get wrong when trying to get their people to actually use these new AI tools?
Ans: A very consistent pattern emerges from the survey across organizations: many are still approaching AI adoption as a technology deployment exercise, when it is fundamentally a behavioral and operating model transformation challenge. The results clearly show that the biggest risk in AI-enabled ERP transformation is not the technology readiness but the widening gap between the leadership intent and adoption at the ground level.
Access equals adoption is one of the most common misconceptions. Organizations are investing in AI capabilities in their ERP systems, clearly articulating a strong transformation vision, and rolling out the tools at scale, but often fail to redesign how work actually gets done inside the enterprise. Three of these gaps are particularly evident in the survey responses. First, AI is not well integrated into everyday workflows. The moment users are asked to leave their core ERP environment to interact with an AI tool, adoption falls off sharply, as employees naturally revert to familiar ways of working. Second, the metrics and incentives for performance do not usually change. Teams are still being measured against legacy process benchmarks and have little incentive to trust or rely on AI-assisted recommendations. Third, training programs are still to feature centric. Users see what technology can do, but they don’t see how this changes decision-making fundamentally, or what behavioral changes they are expected to make.
The problem then is not employee resistance – it is the lack of contextual integration.The organizations that are achieving better results are taking a different approach to the transformation. They’re embedding AI directly in the flow of work, aligning KPIs to AI-enabled outcomes, and empowering teams to make better decisions—not just use new tools. Organizations don’t fail because people don’t want to adopt AI at the end of the day. They struggle because they don’t reengineer the work itself in a way that makes AI a natural and necessary part of how decisions are made and work gets done.
Q3. Only 59% of companies say they have a proper governance framework for AI in ERP. What happens to the other 41% if something goes wrong with their AI system, a bad output, a compliance issue, or a data problem?
Ans : When something goes wrong in an AI-enabled ERP environment, whether it is an inaccurate recommendation, a compliance lapse, or a data integrity issue, the consequences rarely remain confined to technology alone. Very quickly, the impact extends into operations, financial performance, regulatory exposure, and ultimately, organizational reputation. Three main issues consistently come up based on the results of the survey. First, there is usually uncertainty regarding accountability. The ambiguity in governance frameworks makes it difficult for organizations to establish ownership for an AI-based decision-making process, whether by the business unit, the IT department, or the software vendor itself. This ambiguity delays reaction time and escalates the problem whenever it arises. Second, the majority of organizations continue to react to AI-related challenges in a piecemeal manner. Due to a lack of established control procedures, audit records, or escalation processes, each issue is handled on a case-by-case basis, raising the risk of repeat incidents and lowering organizational resilience. Third, the risks related to regulation and compliance are rising significantly. The integration of AI functions into financial transactions, procurement, and customer record management in ERP systems increases the possibility that even a minor oversight will result in audit findings, regulatory investigations, financial consequences, or a loss of reputation. One of the most prominent lessons to emerge from this survey is that there is a disparity between governance and adoption levels. The majority of organizations are quickly scaling up their use of AI capabilities, yet the frameworks that would support them in doing so, such as AI model validation, explainability, data governance, and oversight by humans, continue to develop. However, the organizations that are further along in their maturity, as evidenced by the results of the survey, have already begun incorporating AI governance into their ERP ecosystem in a more formalized fashion. This means establishing ownership, creating approval processes, continuously monitoring AI models, and categorizing AI use cases based on risk. Therefore, for those organizations classified within this group (41%), it is not a matter of whether an AI project will fail; it is only a matter of time.
Q4. According to the report, seventy-five percent of leaders expect to see clear returns within 12 to 24 months. Is that a realistic timeline?
Ans: In the context of the survey findings, the expectation of achieving returns from AI-enabled ERP initiatives within a 12–24-month window is far more realistic than many organizations initially assume, provided the approach remains focused, disciplined, and execution-led. The messages coming out of the research are that very few companies are starting their journey from a completely greenfield position. Many already have ERP environments and are adding AI on top of existing business processes, thereby greatly speeding up the time-to-value process. The level of certainty regarding the time frame is also highly dependent on the types of applications that are being considered. Instead of taking a more experimental approach toward AI, many are focusing on high-impact areas like demand forecasting, inventory management, finance automation, and procurement analytics, which have concrete impacts and tangible value. On the other hand, the report makes one significant differentiation. Although the timeline is quite realistic to achieve positive returns, it does not mean that transformation maturity will occur. There is no correlation between early ROI and transformation. Organizations that manage to show better results at the beginning often possess certain qualities. They invest in use case implementation and not in experimenting with everything possible, rely on existing foundations with respect to data and processes, and have clear ownership and alignment from the business side. In other words, they focus on operational benefits and do not try just to implement new technologies. If they see AI as a long-term and big transformation project, it means that problems are ahead of time. Thus, the conclusion that can be made from this survey is the following. It is definitely possible to generate returns within 12-24 months in the age of embedded AI in ERP systems. However, under the condition of effective and focused implementation. Otherwise, such projects will become nothing more than failure stories for their initiators.
