Capgemini Report: Generative and agentic AI deliver tangible ROI

A new report from the Capgemini Research Institute, titled “AI in Action: How Gen AI and Agentic AI Redefine Business Operations”, reveals that artificial intelligence is now delivering measurable returns on investment (ROI), with organisations reporting an average return of nearly 1.7 times. The report emphasises that this success is paving the way for broader implementation of agentic AI systems. Among early adopters of generative AI (Gen AI), around 30% have already integrated AI agents into their operations. Agentic AI projects are expected to increase by 48% by the end of 2025.

The research finds that one in five organisations already use AI agents or multi-agent systems, with Gen AI and agentic AI contributing to notable cost savings and operational efficiencies across key business functions.

While some organisations initially expressed concerns about realising ROI from large-scale AI and Gen AI deployments, the report indicates that these doubts are diminishing rapidly. As enterprises witness substantial returns—an average ROI of 1.7 times—they are accelerating their investment in Gen AI. In fact, 62% of surveyed organisations have increased their Gen AI investment this year compared to last.

“Gen AI and agentic AI have the potential to revolutionise business services, enabling a shift away from traditional cost-focused models towards AI-driven, value-led operations,” said Oliver Pfeil, CEO of Business Services at Capgemini and Member of the Group Executive Committee. “Those adopting an integrated approach with data and AI at the core will be best positioned to become truly connected, frictionless enterprises. While adoption of AI agents is growing, organisations still face several barriers to scaling. A pragmatic approach, grounded in trust and a solid data foundation, will be essential to turning business services into a strategic engine of growth,” Oliver Pfeil added.

Gen AI adoption lays the groundwork for agentic AI

Over the next three years, Gen AI is expected to improve key performance metrics including insight accuracy, productivity, time to market, and both customer and employee experience. As its value becomes clearer, adoption is rising rapidly: 36% of organisations are now using Gen AI, up from just 20% last year.

Of those that have adopted Gen AI whether at a limited or full scale approximately 30% have already integrated AI agents into their processes. The number of AI agent initiatives per organisation is projected to increase by 48% in 2025.

The report notes that AI agents and multi-agent systems are already delivering significant results across business functions: reducing errors, boosting customer satisfaction, enhancing efficiency, and cutting operational costs. The top five sectors leading in AI agent adoption are high tech, industrial manufacturing, consumer products, energy & utilities, and pharmaceuticals & healthcare.

Leadership and Workforce Transformation Critical for ROI

To maximise ROI from Gen AI, the report underscores the need for strong leadership, robust governance, and enterprise-wide AI readiness. Organisations with such foundations achieve returns 45% faster than their peers. However, only one in three leaders currently acts as a strong advocate for Gen AI within their organisation.

The report also stresses the importance of workforce transformation in realising the full potential of AI. Over the past two years, organisations that have implemented AI and automation use cases have automated 30% of operational tasks and expect to automate more in the near future. As job responsibilities evolve, upskilling, reskilling, and role transitions will become increasingly important. Nearly two-thirds of employees anticipate changes to their job descriptions by 2028. With employee interaction with AI agents expected to increase significantly, training and development will be vital to foster effective human-AI collaboration.

India insights

According to the survey, around 62% of Indian organisations globally have increased their investment in generative AI (Gen AI). In India, this figure is even higher, with 79% of surveyed executives reporting an increase in Gen AI investment compared to the previous year. Among these Indian executives, 49% indicated that the rise in Gen AI investment came from both reallocated existing budgets and additional funding, while 27% cited reallocation alone, and 24% noted that the increase was funded entirely through new budgets.

Globally, three in four executives prefer proprietary models for AI implementation. In India, 36% of surveyed executives stated they plan to implement AI and Gen AI models primarily through proprietary models offered by hyperscalers such as Microsoft Copilot and Google Gemini. Additionally, 33% intend to use proprietary models developed by niche players like OpenAI’s GPT-4 and Anthropic’s Claude.

The top reasons cited by Indian executives for choosing proprietary models include performance, with 77% highlighting the high accuracy and optimised infrastructure; security and compliance, mentioned by 74% compared to 66% globally, due to robust security features and adherence to industry standards; integration, noted by 69% for the seamless alignment with existing enterprise systems; and support and maintenance, identified by 61% for access to dedicated support and regular updates.

In terms of Gen AI maturity, global adoption increased 1.8 times year-on-year in 2025. Among Indian respondents, 37% revealed their organisation is currently building a Gen AI strategy without any active projects, 29% have implemented Gen AI capabilities in select functions or locations, and 20% have begun exploring its potential. Furthermore, 25% of Indian executives said their organisations plan to adopt AI agents or agentic AI systems within the next two to three years, while 23% have already deployed them.

The report also found that global AI leaders achieve ROI 45% faster than others. In India, 39% of executives noted that their leadership is a strong advocate of Gen AI, while 19% described their leadership’s stance as a “wait-and-watch” approach. As AI and Gen AI increasingly impact the workforce, Indian executives anticipate significant shifts in how employees interact with these technologies. By 2027–2028, 58% of the workforce is expected to work directly with AI agents or agentic AI systems, such as by providing input or monitoring outputs. This figure drops to 43% by 2026, while 25% of the workforce is already engaging with such systems.

Regarding access to AI tools, 68% of the Indian workforce is expected to have access to their organisation’s approved AI and Gen AI tools and applications by 2027–2028, slightly higher than the global average of 64%. By 2026, 55% of the workforce will have such access, while 45% already do. Training is also a key focus: 68% of Indian executives expect their workforce to be trained on AI and Gen AI technologies by 2027–2028 (compared to 65% globally), 56% by 2026, and 43% reported that training has already taken place.

Methodology

The Capgemini Research Institute surveyed 1,607 executives from organisations with annual global revenues of at least USD1 billion. All respondents were directly responsible for or involved in AI and Gen AI initiatives related to business operations. Participants represented functions such as supply chain and procurement, finance and accounting, HR, customer operations, AI strategy, application development, and AI ethics and compliance. The survey spanned 15 countries and 13 industries. In addition, Capgemini conducted in-depth interviews with 15 senior executives leading AI and business operations initiatives in their respective organisations.