A poll of leading companies at the Capgemini Research Institute found that while using artificial intelligence (AI) to automate repetitive business tasks offers significant savings, the tasks that are improved are relatively simple. A considerable number of business leaders surveyed say they prefer their own AI implementations over open source alternatives.
A survey of 1,607 executives from organizations with global revenues of at least $1 billion shows that business leaders have reduced customer operation costs by 40% thanks to AI and Generated AI (GENAI). Voting executives also saw a 26% reduction in operating costs for people and 24% reduced financial and accounting expenses. Respondents also achieved a 21% reduction in supply chain and procurement costs.
As an illustration of the potential of the industry's agent AI, Capgemini highlighted Yum Brands, the parent company of Taco Bell, which operates 60,000 restaurants around the world. The company proposed to track crew attendance and plan shift patterns by introducing AI-powered restaurant managers and tailored opening hours to match market conditions.
Although such examples illustrate the possibilities of AI and genai to improve business efficiency, the survey found that a significant portion of the profits reported by respondents tend to be related to automating simple, repetitive tasks. According to Capgemini, this suggests that the use of AI and Genai in voted executives represents early stage efficiency rather than long-term transformational impacts.
Savings need to be balanced with the cost of running an AI system. The Capgemini Research Institute noted that the price of inquiries for trained models has dropped dramatically. For example, Openai's GPT 3.5 fell from $20 per $100 to $0.07 per million, while GPT-4 fell from $15 to $0.12 per year.
Methods such as model pruning, quantization, and distillation can be used to reduce the size and complexity of AI models. As the Capgemini Research Institute notes, these optimized models need to have fewer computational resources to reduce inference costs. In addition to more efficient algorithms, the Capgemini Research Institute stated that efficient hardware utilization, batch processing of inference requests, dynamic scaling that adjusts the number of computing resources based on current demand, and energy-efficient algorithms can significantly reduce the power consumption of AI models.
However, while open source models such as DeepSeek have been shown to reduce computational costs by 11 times without compromising performance, it has been shown that they can address the sophisticated hardware bottlenecks faced by many organizations, the votes have shown that executives are less enthusiastic about open source AI compared to proprietary AI models.
Despite the increased performance and cost benefits of open source AI models, Capgemini reported that a considerable majority of executives continue to support the implementation of their own AI. Three-quarters of the executives surveyed prefer their own models, with 43% choosing one developed by Hyperscalar and another third-selection model from niche providers.
Capgemini has found that its unique model and preferences for AI systems are particularly strong among organizations that have expanded their investment in AI and Genai. This is suggested by the authors of the report, and shows a clear trend towards trusted enterprise-grade AI products that offer robust support, security and integration capabilities.
Findings published in Capgemini Research Institute's AI Action The report identifies many trade-offs that will curb companies' adoption of open source models due to trade-offs that IT and business leaders need to make. These include the need to improve technical expertise, potential exposure to security vulnerabilities, and the reliance on community-driven support that can affect update cycles and documentation quality.
“We are pleased to announce that Capgemini's business services company Oliver Pfeil said: “Genai and Agent AI can truly transform business services, enabling a transition from traditional cost-focused models to AI-enabled and value-driven business.
However, he said the study suggests that organizations face many barriers to reducing the deployment of AI agents. “Take a practical approach, promote trust in AI and create a strong data foundation will go a long way in transforming business services into a strategic powerhouse and promoting any company,” he added.
