- In a Deloitte survey, 26% of managers said their organizations seriously explored autonomous agents.
- ServiceNow, SAP and Salesforce are among the companies that have made their debut for AI agents to perform work tasks.
- This article is part of “AI in action”, a series exploring how companies are implementing AI innovations.
When customers of the cloud service -based software provider contact the company’s customer support center, 80% of cases – in the form of calls and chat messages – are treated without any human intervention.
Instead, the company is based on analytical and generative artificial intelligence – in the form of AI agents – to answer current questions from customers.
Chris Bedi, director of the ServiceNow customers and business advisor, said employees still manage a five out of five requests.
They obtain a new AI agency medium, which can automate tasks such as writing an e-mail to answer a customer. The workers remain in the loop for a final signature before agent-to-actions are executed. The combination of human workers and agentic AI has reduced the time to manage the most complex cases of 52% in a period of two weeks, said ServiceNow.
The co -founder of Openai, Sam Altman and other leading technologists, said that 2025 will be the year when AI agents “join” the job market.
In addition to serviceNow, software developers such as Salesforce and SAP have deployed their own agents-AI platforms. These can carry out workplace tasks such as the treatment of customer bills, the provision of customer support to customers and the writing of emails. The Intuit corporate software giant, which has Turbotax and QuickBooks, began to deploy agent-oa capabilities in December.
Humans remain mainly in the loop for the moment, but the sellers predict that this technology will become entirely autonomous. Multi-agent systems, where two or more AI agents collaborate to finish work, proliferate.
“Agents are the next level of understanding around the way you apply AI,” said Jim Rowan, AI head of the Deloitte consulting company. “He can perform actions for you.”
In a recent Deloitte survey of 2,773 business leaders, 26% of respondents said their organizations explored autonomous agents in a “large or very large measure”.
Why AI agents have become the new generative AI objective
During the first two years of the generative -AI Boom – which started after the beginnings of the OpenAi Chatgpt at the end of 2022 – most of the companies that adopted technology widened it for electric chatbots and full routine tasks such as writing meetings. AI agents represent an evolution of generative-ai technology, designed to carry out tasks independently, although most are always closely monitored by workers.
AI of agentics “actually has unique skills around reasoning, planning and orchestration,” Bedi told Business Insider. “These agents can collaborate with each other and start trying the promise of work independently.”
BUZZ for AI agents entered high speed after Salesforce made its agentforce debut in September to automate the tasks of customer support, sales and marketing. The company said that it would deploy 1 billion agents to customers by the end of this year. The company also indicated that more than 340,000 of its customer support issues had been answered independently with Agentforce.
ServiceNow estimates that the company’s AI agents, already deployed in various parts of the company, such as customer service, human resources and IT, lead around $ 325 million in annualized value by strengthening the productivity of the workplace of 20%. ServiceNow says that the work supported by AI-AGENT saves 400,000 hours of work per year.
However, technological companies are in the early stages of their agent-like development. Many determine the processes they can fully automate with technology. Consequently, business leaders implementing an agentic AI form their workers to collaborate – and to provide comments on – their new “colleagues”.
AI agents are often developed as workers’ collaboration tools
John Kucera, vice-president director of product management at Salesforce, recommended that companies be transparent on the work that AI agents can manage and what will remain with workers. He added that companies should be clear about what an AI agent really is, saying that not all agent systems are equal.
“There are a lot of false agents there,” said Kucera. “It is only an agent when you take a request and the agent determines what to do, then what data put.”
While surveys frequently note that many workers fear that AI will replace them, technologists say that AI agents will not replace people but take responsibility for banal tasks.
“These agents will help me do my job, but at no time will they do something to me that I do not know,” said Walter Sun, the AI world chief at SAP, who sells software for financial needs, the supply chain and other business management needs. “The most important thing is that employees still control.”
How companies adapt AI agents with employee comments
To ensure that workers have a voice in the way AI agents are developed, SAP encourages employees in its various business sectors – including the management of Voyages and Success Factors, which provides HR, payroll and talent management software – to use an internal online form to contact the AI team and offer cases of use of the agency.
At Intuit, financial assistance fed by AI Intuit Assist can charge companies 45% more quickly by detecting past invoices and automatically reducing a personalized reminder note. After a business owner approved the language of the ticket and sent it, it is paid, on average, five days earlier than with a process only, said Intuits Assist.
But before the Inuit helped, humans have the last word. “What we are trying to do is having a good human-automation interaction,” said Ashok Srivastava, Director of Intuit.
Intuit has adopted an Ai Robust training program, focused on responsible AI and what technology can and cannot do, and has built a “sandbox” called peopletudio which allows employees to interact with large models of language in a secure environment. The company has also developed educational programs adapted to senior executives, administrators and engineers. “It is very omnipresent throughout the business,” added Srivastava.
Asana, who manufactures work management software, launched AI agents in October, focusing on some functions, including marketing, IT, HR and research and development. Rather than following a specific number of actions that the agent AI takes over, Asana monitors the types of work that can be automated, eliminating the chore of work occupied to allow employees to focus on more complex tasks.
The company also closely monitors the tasks that AI systems are mistaken compared to people. In cybersecurity, human errors tend to occur later in the day, when workers are tired after a long change. The AI is not tired, but it is sensitive to hallucinations – or when a model of AI generates a misleading response or false information, but which nevertheless presents it as a fact. For example, ASANA’s IA agent could answer certain questions by suggesting tasks that are, in reality, nonexistent to a particular workflow.
“The types of errors we see are different, so the way we rely on them must be different,” said Asana information security director Sean Cassidy. He said the company performs automated tests to detect hallucinations and improve the product when they occur.
For AI agents to create a convincing return on investment for the companies that implement them, said Rowan de Deloitte, workers should be responsible for frequently checking simple automated tasks before agent AI actions are taken.
If companies want to see great yields on their original IA investment, they must place AI at the center of their work model, then consider how humans will engage with work, said Rowan. Otherwise, “the savings will really not be there,” he added.
Bedi’s Bedi de ServiceNow said that the success of AI agents depends on companies that nail three factors: new agent capabilities should be developed for each department and its specific needs, unique training plans should be designed for each part of the company – such as finance, marketing and sales – and value and return to agent workflows should be monitored closely.
“Companies that combine these three ingredients will have a competitive advantage,” said Bedi.
businessinsider
