娇色导航

Agent-based AI is fast becoming the hottest technology in IT. Recognized by Gartner as climbing steadily toward the peak of inflated expectations, agentic AI may take between five and 10 years to reach peak productivity, according to the research firm.  

IDC shares this view, pointing out that AI assistants currently dominate the market but autonomous agents will be the ones that triumph in the medium to long term.  that by the beginning of the next decade, AI agents will replace entire applications: Why use CRM software when you can rely on a fleet of CRM agents?

Interest from IT leaders is undeniable: 82% of IT leaders say they’ll use agents as digital team members to expand workforce capacity in the next 12 to 18 months, according to a  focused on the emergence of a new organizational model with AI integration at its core. The impact will be such that Microsoft expects a new professional role to emerge: the agent manager.  

Despite near unanimous belief in the future importance of agent-based AI, development today remains in the early stages. According to a , three sectors are already experiencing the first wave of agentic AI’s impact: finance, retail, and healthcare.

Gonzalo Goñi, director of solutions engineering at Salesforce, developer of the Agentforce platform, says, “Agent adoption is booming, and for now, what we’re seeing is a clear and progressive acceleration. Organizations are rapidly moving from experimentation to production and are increasingly integrating these agents as a structural part of their operations.”  

Here, IT leaders from finance, retail, and healthcare lend insights into what their organizations are doing with AI agents today — and where they see the technology taking their organizations and industries in the future.

Salesforce.

Integration in retail

Goñi sees demand for AI agents rising across all sectors, but the retail industry is where the action on agentic AI has been the most dynamic. Here, Goñi explains, use of agentic AI has been focused on “optimizing the customer experience and logistics operations.”

Fernando Herranz, director of digital experience and ecommerce at Leroy Merlin Spain, agrees, seeing the promise of “a progressive but constant integration” of agentic AI with the company’s comprehensive omnichannel model, with beneficial impacts for both Leroy Merlin’s clientele and its teams.

“AI is allowing us to move toward a more fluid and personalized relationship with the customer, and we do this by combining strategic vision with controlled and measurable applications,” he says. The company is exploring use of agentic AI, for example, in store automation processes, digital content generation, and personalized support, both online and in physical stores.  

Herranz and his team are working toward an “orderly and focused rollout on specific use cases with measurable impact.” Initiatives already implemented include the company’s 24/7 virtual assistant based on generative AI and its Greeter AI pilot project for physical stores.  

One of Leroy Merlin’s main challenges has been integrating the technology with its omnichannel ecosystem to enhance real-time, scaled, and highly personalized operations. This “requires a solid, interoperable, and constantly evolving technological infrastructure,” ensuring data protection and algorithmic transparency, the “key to maintaining customer trust,” he says.

Looking ahead, Herranz points to “a significant expansion of the virtual assistant, which will evolve to manage incidents, perform after-sales functions, and support customers throughout their entire relationship with the brand.” 

Leroy Merlin.

Another retail company that is investing significantly in AI is Carrefour, which launched its ai.Carrefour project in Spain in April. The central objective of this initiative is to facilitate the digital transformation of teams to improve customer service, explains Francisco Escalona, ​​CDO of Carrefour Spain.

Among the solutions Escalona and his team have implemented is Super Agente Uno.ai, a virtual assistant “designed to support strategic decision-making and optimize the company’s operational processes, from product management and logistics to store efficiency,” he says. With the agent, they can also detect new trends and strategies in healthy and sustainable eating or improve customer experience and loyalty through data analytics.  

Escalona also highlights Carrefour’s ClubAI, “the first virtual assistant with artificial intelligence for a loyalty program in Spain.” It analyzes each member’s preferences and purchasing habits “to offer a unique and hyper-personalized experience thanks to its ability to interact with [customers],” with recommendations and content suggestions that add value, such as recipes tailored to customers’ preferences, or information about Carrefour stores. The strategy for deploying these tools has recently included the creation of the chain’s AI Advisory Council, “which demonstrates Carrefour’s commitment to this technology.”  

Both the CDO of Carrefour and the director of Leroy Merlin agree on one of the main benefits of agentic AI for their shared industry. “It’s an opportunity to enhance talent and build a more enriching future thanks to training in tools that help us with repetitive tasks so we can focus on higher-value-added activities, such as creativity, innovation, and personalized customer service,” Escalona emphasizes.

Herranz shares this mission of “freeing the team from repetitive tasks that could be automated and allowing them to focus on what adds the most value.” 

Carrefour.

Agents for the financial sector 

In the financial services sector, large companies are already deploying agents to automate critical processes, improve customer service, and strengthen fraud prevention.

At Ibercaja, agent-based AI is seen as “a sophisticated extension of generative AI,” explains Nacho Torre, director of strategy, transformation, and data at the financial services company. Torre and his team are still in a very early phase with agentic AI, as the company navigates two aspects that must be further refined in order for its use of AI agents to expand: one regulatory, and the other operational.  

The company is currently in an “exploration and preparation phase,” Torre explains, “finalizing the identification and classification of existing AI systems, assessing their risks, and defining a governance model that will allow for their formal control starting in the second half of the year.”

This framework, Torre says, will include agents that will eventually be developed to cover “simple but high-impact tasks.” Ibercaja is already considering ideas such as an agent that automatically connects to the Bank of Spain website every morning to download the legal interest rate on money.

“This type of automation, although basic, represents a clear and replicable use case,” Torre says. In the medium term, the firm plans to develop “multi-agent platforms, capable of coordinating multiple tasks and collaborating with each other, which would open the door to more complex and strategic automation.” 

Even at this early stage, Torre delves into the challenges of implementing agentic AI. From a technical perspective, he prioritizes both security and privacy, as well as adapting to a dynamic environment.

“It is essential to build flexible and scalable technological platforms capable of quickly adapting to market changes without compromising security or governance,” he says.

From an organizational perspective, Torre emphasizes the need to develop technology that is “extremely easy to use and geared toward specific use cases that provide real value.” 

Ibercaja.

Agentic AI in healthcare 

Autonomous agents also have enormous potential in healthcare as well, Salesforce’s Goñi contends. “The use of agents can alleviate the administrative burden on healthcare professionals and improve patients’ access to information,” he says.

For Dr. Ramón Puchades, an internal medicine specialist and coordinator of the digital medicine group of the Spanish Society of Internal Medicine (SEMI), integration of AI into the healthcare organization’s operations is still in its early stages. “Even in a phase that we could define as an AI bubble,” he says.  

Puchades emphasizes that studies and research are still needed to validate what are the most plausible use cases, but he does distinguish three fields of application in which both agents and generative AI are emerging: clinical practice, research, and teaching.

Regarding clinical practice, Puchades highlights AI agents’ value as tools to support diagnosis, patient monitoring, and clinical progress. In teaching, it can be used for “slightly more active” learning, while in research, he integrates it for scientific literature searches, writing support, or even in clinical trial recruitment.

To these, he adds its use in management tasks, “to try to organize activities and establish patient flow.”

“But all of this is still at a very early, very initial stage,” he reiterates. “It’s very important to establish evidence and quality filters, because they are still tools,” he adds. “It needs validation, evidence, and quality when it comes to putting it into practice, because it’s above all about providing benefits to the patient without risks.”