It started with chatbots. 

When AI first swept through the labor market, most people interacted with it through a simple text interface—logging into the Claude website, having a conversation, and using the output in their work. But as the technology has advanced, agentic AI—software that doesn't just answer questions but takes actions on its own—has emerged as the new frontier where AI and the labor market meet.

For our research in this year’s Stanford AI Index Report, Lightcast added “Agentic AI” as a new cluster of AI-related skills. This list of skills is used by us and other leading researchers to track how AI is spreading throughout the labor market—and specific to Agentic AI, we can see that demand is rising fast. (By contrast, postings that mention “ChatGPT” and “Chatbot” actually saw a decrease year-over-year.)

By looking at the overall demand for agentic AI, the employers hiring for these jobs the most, and what kind of occupations are adopting it most, we can see the ways that it is reshaping the global economy. 

Demand Is Exploding for Agentic AI Jobs

Employers posted over 86,000 jobs mentioning agentic AI skills in 2025, and mentions of related skills in postings jumped more than 280% over the past year. The median salary, $153,000, beats $141,000 for AI postings overall and $125,000 for IT postings generally—a premium that signals employers see this as a high priority and an investment in the future of their workforces.

The World’s Biggest Companies Post the Most Agentic AI Jobs

Fortune 1000 firms account for roughly 30% of agentic AI postings, more than double their 12% share of postings overall. Fortune 100 companies alone make up nearly half of that—about 13% of all agentic postings—despite being a sliver of the labor market. Amazon and Microsoft sit in the top 10 hiring companies, alongside consultancies like KPMG and Accenture, who are building agentic capabilities to sell, not just to use.

Agentic AI Workers Are Building Them, Not Using Them

Engineering roles—including AI engineers, software developers, machine learning engineers, NLP engineers—dominate the occupations that mention agentic AI skills. Six of the top 10 occupations are technical. That mirrors the early arc of generative AI hiring, which also started in engineering before spreading into marketing, operations, and beyond.

All of these jobs have been around for years. Agentic AI is being built by the same types of engineers that these big companies already employ—not by a brand-new worker with the title of “Agent Builder” or something similar. While some postings for completely new titles like “prompt engineer” or “AI orchestrator” do exist, they’re the exception. As a general rule, agentic AI is being built within the existing framework of jobs. 

At Lightcast, we often talk about taking a skills-based approach to understanding the labor market, and this illustrates why. If you were looking for agentic AI only in job titles, you would barely see any. By looking at the whole picture, and especially by using skills, we can see that there are nearly 90,000 of these jobs. 

We see this illustrated with AI jobs more broadly: In Q1 2026, AI occupations accounted for just 0.3% of job postings, representing only 7% of all AI adoption. The remaining 93% of AI adoption comes from existing occupations that have incorporated AI skills into everyday work.

There’s a very real chance that eventually, agentic AI titles will become commonplace—before “Social Media Manager” was an occupation, “social media” was a skill in marketing roles. But the odds are just as good that agents fold into existing jobs. Email was once a game-changing skill, but today, it’s used by nearly everyone, not siloed off to a specific job like “Email Writer.” 

Whatever path agentic AI takes, the underlying shifts in jobs, skills, occupations, and organizations will show up in the data long before they show up in the headlines. Labor market intelligence gives employers, educators, and policymakers a way to see that shift coming—and to plan for it instead of reacting to it.

Thanks for reading On The Job. Be sure to catch up on our past issues, including "Oil, Energy Independence, and the Workforce Gap” "Six Dos and Don'ts for AI" and "The Next Great Resignation." You can also subscribe here. We’ll see you next time.