The CEO of Upwork: AI is Creating More Jobs Than it Kills

AI is actually creating more opportunities and jobs than it is destroying by far, and we see this across segments of our platform, which we are happy to unpack.

Thanks for having me, Jeff. My background is a little unusual. I grew up in Kathmandu, Nepal, after my parents moved there for international development work. They instilled in me a lesson that it's worthwhile to spend your life helping other people. That experience shaped my worldview and led me to join Upwork in 2011, a company that brings opportunity to people all over the world who have incredible skills but cannot access opportunities in their local market. Our platform connects them to employers looking for great talent, creating opportunity for both sides.

Our business processes about $4 billion of client spend every year, passing to talent in 180 countries. We have more than 130 categories of work, from design to AI development. In the US alone, freelancing is a $1.5 trillion contributor, with about 28% of Americans freelancing. We have a canary in the coal mine effect, seeing trends before they ripple through the economy.

We've gone through a crawl-walk-run approach. We started dabbling, then realized we needed to make it programmatic. We established a monthly AI business review where every department came in with a two-year vision. That gave leaders a stake in the ground, and we have monthly check-ins to evaluate what's working.

Some of the most exciting ones were in marketing and product engineering. In marketing, they envisioned a connected tool chain around AI that could sense market conversations and create content in real time. In product engineering, they talked about a radically different approach where one person could do engineering, design, coding, and customer requirements, powered by AI tools. That vision is now becoming a reality.

I knew from day one that the AI agenda for Upwork had to be my agenda. After ChatGPT launched, we had a debate in the executive team about whether to change our roadmap. I felt strongly we had to lead, not watch. So we changed our product roadmap and started with an isolated team reinventing our platform. I showed up every week at our all-hands meeting with a message that this technology is exciting, not threatening. We celebrated people embracing it and talked about mistakes too. That created a culture of peer-to-peer learning and reduced fear.

I heard a story of two founders in Germany who did an A/B test: one kept running the business, the other reimagined it with AI. We dedicated a small tiger team to develop a new vision. They came back with a prototype that brought tears to my eyes. They reversed our problem statement and created an AI agent, UMA, that helps customers before they even know who to hire. UMA now writes 70% of new client job posts and helps project manage entire efforts. More components are coming in 2026.

The secret sauce is reinventing, not retrofitting. Many businesses try to sprinkle AI fairy dust on existing solutions. We had to rethink the entire user experience, centering AI in the experience. Also, you have to change the people, not just the technology. Give them new skills, redefine their roles. When you look at talent in conjunction with AI, you get real benefits.

It started with culture and setting goals. We told the company we are reinventing, going back to startup mode. We gave everyone permission to move faster, try things and fail. That cultural layer, combined with specific goals around product and operations, allowed us to achieve what we did. We moved at the speed of a startup and the scale of a market leader.

We have a unique vantage point because we see the freelance economy, which ADP doesn't capture. Macro has played a bigger role than AI in labor market softness. There's a misattribution that employers aren't hiring because of AI, but we're actually seeing AI creating more job opportunities. Employers are hiring more individuals with AI skills. AI is creating more opportunities and jobs than it is destroying by far.

We see two trends. One is the fractionalization of work. Traditional full-time jobs are being broken apart, with parts done by AI and residual work done by humans, but not as full-time jobs. Those opportunities are coming into the freelance economy, where flexible workers take on those pieces.

They don't look like the traditional full-time job. So that is definitely one thematic that is starting to emerge. The other one is a lot of companies as they're tackling these AI projects and their AI ambitions they don't have the people in their team to do that work either because of capacity like people are just busy running the existing business and executing things or because of skills and they don't have specific skills needed. So that's creating net new work opportunities where they're trying to build something out. They come into a place like Upwork. They need to assemble a team or bring in experts to do that work and then they build from there. And a great example of this is a company like Grammarly. Grammarly was building an AI writing assistant for their customers. They realized it wasn't just the AI tool that could deliver the value. They actually needed human editors and writers working alongside the AI system to deliver that product. So now they have hundreds of writers and editors working alongside AI in a fractional freelance capacity. And that's just net new work that wasn't happening before because of this new product that they launched. So we're seeing both of those trends, which I think are why we're going to see over the next few years a shift actually from full-time work to freelance work accelerating because the economy is going to keep changing. The nature of work is going to keep changing and freelance and flexible work is actually better suited to a lot of the work that's being created because of these AI implementations.

So the AI category in our platform is about $300 million of spend as of last year and it's growing at about 53% year-over-year, so it's a fast growing category. You're right, things like prompt engineering showed up a few years ago more prominently, and actually they're still enduring. But what is changing is businesses are getting more sophisticated about what they're looking for because, to your example, everyone thought it was a prompt engineer. Now they're realizing, okay, actually I need something else or something more, something different. And the requirements are getting more specific. Specifically, people are looking for hiring folks who have technical skills but also have specific human skills like problem solving, judgment, creativity, the ability to put systems together. I was literally reading a client job description this morning. They were like, 'I don't want just a prompt engineer. I want someone who can design a system and get these AI systems talking to each other.' Yeah. And it's really about this orchestration layer. So more and more companies are looking for these AI generalists who know enough about how AI systems work to make them useful, but also need to put them together and orchestrate them in ways that no single tool or someone who just knows about one tool can necessarily do very effectively.

You know, the debate often feels really binary, Jeff, and I think that's a mistake. I think that's a disservice to all of us. So you hear factions out there saying jobs are going to be wiped out, software development is going to be gone in 12 months, whatever these prognostications are that are very dire. And then there's kind of the Pollyannaish like, 'No, but this is just going to be net new.' The reality is definitely in between. There are going to be winners and losers, there's no doubt. But the real question is the pace at which we can help reskill and transition people from the old to the new, and the pace at which businesses can transition, because I think there's a confusion that happens between the pace of technology change and the pace of business adoption.

Yes. The tech is moving super fast. Capabilities are enhancing. But I talk to these companies every day and it's hard, like we were talking about. It's hard to make these changes. It's hard to evolve your systems. It's hard to get the right talent. You know, we're democratizing that access to AI talent for small businesses, but there's still work that they've got to do to go down this path. And so, businesses can only move so fast. And I think that is actually a blessing because it creates a little bit more time and a little bit more buffer for workers to reskill, to learn new things, to adapt. And we see that happening all across our platform where, yes, there are AI categories of work that are growing, but we look at every category on our platform and it is becoming AI-powered, AI-enhanced. So it's a freelancer and an AI agent or AI tools working together to create way more value than what they could do individually, either one of them. In fact, we've studied this at length. We have an index we put out, we call it the Happy Index, the Human and AI Productivity Index, and it actually shows that AI tools can do very little on their own, but when you put humans working with them, their results increase by 70% or more. So, I think this binary narrative of AI winners, AI losers, and nothing in between is not true. In fact, every job is going to be AI-enhanced, every worker is going to be AI-powered, but there is going to be net new opportunity that's very exciting, and we just need to get people over that bridge and cross that hurdle as it's coming.