Artificial intelligence systems capable of planning and executing complex tasks are beginning to replace entire corporate departments, as companies move beyond experimentation and deploy “agentic AI” in real operations, Anand Sahay, global CEO at Xebia, told TNIE.
Unlike earlier AI tools that mainly helped employees work faster, agentic systems can autonomously perform complex operational tasks that previously required teams of people.
Sahay cited the example of a large global apparel retailer that has automated a major part of its supply-chain ordering process.
The company sources clothing components such as fabric, buttons and zippers from multiple countries, including China, India and Bangladesh, and must decide which factories to use, how much to order, and whether suppliers meet regulatory rules.
Previously, teams of employees handled these decisions by analysing supplier reliability, compliance restrictions and delivery timelines.
That function has now been largely automated.
“Our AI is able to make a better determination and calculation of which factory is the best for this, how much from where, equaling the total quantity you need,” he said.
Human involvement has been reduced to minimal oversight. “You can imagine its departments, a very small number of people who are now just brought as a human in the loop to say this is what is getting ordered, any questions, one button.”
The impact on staffing has been significant. “So productivity is reaching another level that way. It's two months with 100 people, now you need one guy and maybe less than three days,” Sahay said.
However, he said many organisations are still struggling to capture similar gains because they apply AI to existing processes instead of redesigning them.
“Tech has become much simpler. Tech is very clear now,” he said. “It is the businesses which have to rethink how they want to run their business.”
As a result, many companies remain stuck in proof-of-concept projects rather than large-scale deployment.
Financial services firms are among the sectors moving quickly, particularly insurance companies, which he said are using AI to automate processes that were previously manual.
But the technology also raises regulatory and legal risks. Banks in particular are concerned that automated lending decisions could introduce unintended bias.
In the United States, financial institutions are worried that errors could emerge years later.
“What if I make a mistake and then we found three years later that for a particular kind of minority, the lending rates were higher by mistake,” he said. “And then I will have a class action suit five years later.”
Despite such concerns, Sahay said the broader impact of AI is likely to be lasting. “This is for real,” he said. “This will fundamentally change things.”