According to the 2026 edition of Salesforce's Agentic Enterprise Index, compiled from real data from its Agentforce platform, the number of AI agents activated per company has (almost) tripled, while the average time required to create them has been reduced by 53%, to just two days. Obviously, this data refers to companies using Agentforce, and therefore represents cases where there is a predisposition to using agentic technology, but it is still interesting data. The report reveals that agents are evolving from simple conversation to direct action: they consult records, apply business rules, and process complex tasks from start to finish. To measure this capability, Salesforce has introduced the concept of the Agentic Work Unit (AWU), which represents a specific task completed by an AI agent. In April 2026, the last month for which data is available, monthly AWU production was growing at a rate of 15%, led by the retail sector. As Salesforce explains, “Consumer-facing sectors, especially those with a high volume of customer interactions, lead AWU production. In the case of retail, agents tend to focus on very specific tasks, averaging one or two actions per agent most of the time, reflecting a high volume of routine work.” The report highlights that retail accounts for 22% of the total monthly AWU volume, with an 18-fold increase between February 2025 and April 2026. Meanwhile, the travel sector represents 10% of the monthly volume and registered a seven-fold increase over the same period. How Agents Are Used in Businesses Customer Service During Peak Season The study identifies two main implementation models: on the one hand, solutions geared towards specific tasks and high volume in the consumer goods sector; on the other, versatile, multi-stage agents in regulated industries such as finance, manufacturing, and healthcare. The versatility of these systems is key, as they can expand their skill set by up to 350% during critical periods, such as holiday shopping in the retail sector. In fact, during the holiday season, retailers that used AI agents saw an 8% year-over-year increase in online sales, compared to 2% for those that did not. One example of the use of this technology is the jewelry company Pandora, which typically experiences a significant increase in customer inquiries during peak shopping seasons, especially Christmas and Valentine's Day. In this case, Pandora implemented Gemma, an AI-powered virtual assistant developed by Agentforce, which it uses to answer common customer questions (from order status and shipment tracking to frequently asked questions about jewelry care). In addition, Gemma offers personalized gift recommendations based on each customer's preferences. During peak season, Gemma handles six out of every ten routine support requests. Salesforce reports that using Gemma for these tasks results in a 10% increase in Net Promoter Score (NPS), allowing agents to focus on more complex and personalized interactions. Employees also use agents. Internal adoption and trust are also growing. During the period analyzed (February 2025–April 2026), employee agent usage skyrocketed: each employee interacted with an agent 300% more often per week. As an example of internal use, Salesforce notes that 83% of its workforce regularly uses Slackbot, saving an average of up to five hours per week per employee. From the customer's perspective, agents are also having a significant impact. Agents autonomously resolved seven out of every ten customer service conversations. Furthermore, recent research cited by Salesforce indicates that 77% of shoppers who interacted with integrated buying agents on websites felt more confident about their purchase.