ABOUT 805,000 students finished college in 2023-2024, according to the Commission on Higher Education. Some have spent the past year looking for work that is not there. So when I sat down with Vaibhav Vora, chief technology officer of Ascendion, at its Makati office this week, that was the question I kept asking. Ascendion calls itself an AI-native, platform-driven engineering services company, and AI is changing the offshore work that has employed Filipinos for two decades. For someone starting out, is that a better job or no job?

Vora’s case starts with a gap. Companies have increased their technology spending by about 8 percent a year since 2022, while productivity has grown by only 2 percent. Those are McKinsey’s US figures, and the firm itself acknowledges that the relationship is not exact. When I asked whether the pattern holds here, Vora said it extends beyond the United States to Europe and the Asia-Pacific region, including the Philippines. His explanation is that most companies leave AI trapped in pilot projects instead of deploying it across the business. Getting it from pilot to production, he says, is the work Ascendion does. He calls the platform AAVA, an in-house system that runs AI agents across the full software development cycle rather than one task at a time.

I asked whether that was a diagnosis or a sales pitch, since fixing old systems is what Ascendion sells. Modernization is as old as the industry, he said, and the difference is the method. He pointed to a US bank overhaul that Ascendion’s AI agents delivered at one-third the cost.

Ascendion has operated in the Philippines since 2013 and has worked with more than 80 local clients, including the global capability centers (GCCs) that multinationals run here. On the jobs question, Vora was direct. “For us, it is not about replacement of jobs,” he said. His clearest example is a customer service operation the company runs for a multinational in the Philippines. Before AI, the average call lasted 20 minutes. With AI tools, he said, that fell to two or three minutes, and headcount did not decline. Agents now see the caller’s history and preferences on one screen, drawing information from what used to be 10 separate systems. The entry-level role does not disappear in his telling. It undergoes, in his word, “remodeling.” Today’s hires need prompt and context engineering skills, along with data science capabilities that the job did not require 25 years ago.

That is the hopeful version. The harder one came from Dr. Paul Roehrig, Ascendion’s chief strategy and marketing officer. People are smart, he said, and they can see that the same work may require fewer hands. “And that may be true.”

What complicates reskilling is where demand comes from. Vora described meeting the head of a GCC that morning, who was under pressure from headquarters to cut costs. AI, he said, was the mechanism for delivering those savings while freeing people for higher-impact work. Ascendion’s fact sheet puts it in plainer terms, promoting work that is “50 percent more economical” and pointing to a project that “freed” 6,000 engineers. Both are real — the retraining and the cost-cutting — and workers are caught in between.

This is not an abstract issue. Two weeks ago, I wrote about entry-level hiring in the business process outsourcing sector tightening as AI absorbs the simple work that once trained new graduates. Earlier, I profiled a Filipino freelance writer whose income dried up when clients shifted to AI. Vora is right that people can retrain. But there are fewer entry-level jobs where they can learn the work in the first place.

So what should a graduate study? Vora, an engineer whose 18-year-old son will start university soon, gives the same answer he gives his child. Build a strong foundation, where analytical reasoning comes from. Then move beyond older programming languages to Python and develop a grounding in data science. Become fluent with tools such as Claude and Microsoft Copilot. He built a program he calls 10X around early-career hires. Nine out of 10 members of that team are entry-level employees, he said, and his best-performing engineer in the United Kingdom graduated only last year. He sees new graduates as the quickest to adopt the tools.

The fear is real, Roehrig said. In the United States, commencement speakers who mention AI have been booed. His answer is that the work waiting to be done — the legacy systems that organizations have not had enough people to modernize — is greater than the work being automated away. He may be right. But he is betting on jobs that do not exist yet, and his company sells the tools meant to create them.

What I can tell graduates is simpler. The rung they are reaching for has moved higher. Vora’s advice to his son is the closest thing to a map: learn the fundamentals, then learn how to ask the machine the right questions. His son will start university with that map in hand. How many of the 805,000 graduates will reach that rung, and what happens to those who do not? Their answers never quite settled the question.