Jensen Huang: Engineers Build Nvidia AI agents

Nvidia AI agents are now the main focus. The company's software engineers are writing less traditional code than ever before, and chief executive Jensen Huang revealed this shift by explaining that his engineering workforce has developed a distinct preference for building these autonomous systems over writing standard Python. But he frames this evolution not as a threat to engineering jobs. It's a major upgrade in the nature of their daily work. During an interview published by the company, Huang made it clear that the transition from manual coding to system design is a welcome change across his technical teams.

The shift from syntax to system design

The transition away from manual coding marks a change in how the company approaches software development. Engineers are stepping into roles that require them to design, instruct, and manage automated systems instead of typing out line after line of Python syntax all day. But Jensen Huang draws a sharp distinction here. He sees coding as a mere task and engineering as a creative craft, and the actual typing part of the job , the mundane work of translating ideas into computer syntax , is being handed over to intelligent systems. It's a big shift.

But how do these systems actually operate? An AI agent differs from standard software in several key ways:

• It breaks down a large, overarching goal into a sequence of smaller, manageable steps.
• It handles each step in turn, allowing the software to plan and act.
• It operates continuously rather than simply responding to a single, isolated prompt.
• It requires continuous evaluation through rigorous benchmarking and safety guardrails.

Engineers can focus on high-level architecture. They're now building the core agentic systems, writing complex benchmarks, and designing the guardrails that keep these automated workers operating safely within their boundaries.

Commercial logic drives the engineering shift

Deploying agents internally

Jensen Huang has spent months sharpening this message on various public stages. It's a test, really. He's pitched this vision at a Carnegie Mellon commencement address and the Computex trade show, where he describes a future in which his company deploys these digital assistants across every single internal division to lift overall productivity. So by using Nvidia AI agents to run its own operations, the firm is attempting to prove the viability of the technology on its own workforce first.

Selling the hardware underneath

There's a clear commercial motive. The company has spent the past year positioning itself as the foundational infrastructure beneath the rapidly growing agent economy. So it's a highly visible advertisement. A corporate workforce that builds and runs autonomous agents is also a workforce that consumes massive amounts of compute power, and by showcasing its own engineers using these tools, it sells the powerful chips and platforms to the rest of the world. But don't miss the strategy.

Employment debates and the future of work

Nvidia's AI agents are spreading fast inside the company. But that's just one part of a huge debate about the future of technology jobs, where prominent leaders like Anthropic's Dario Amodei and Amazon's Andy Jassy have warned that intelligent software could eventually wipe out large numbers of entry-level and administrative roles. Huang rejects this pessimistic outlook entirely.

The amount of work we've got to do to bring AI into the world is really quite incredible. So it's creating a whole bunch of jobs. And my software engineers love this.

, Jensen Huang, Nvidia Chief Executive

He's adamant: the technology creates roles. But he doesn't believe it erases them. In his view, this technological shift's immediate effect is an enormous number of new jobs, which offers a prime opportunity for industrial modernization across the board. Other firms report slower agent progress than the hype suggests. So Huang stays highly optimistic.

The reality of the new engineering craft

Whether this optimistic view holds true across the wider global economy is a separate question from what's currently happening inside the chip giant. But inside the company, his highly skilled engineers are still fully employed. His argument is simple. Giving talented software professionals better tools allows them to eliminate the most tedious parts of their daily workload, and it's a strategy that's already proving effective in boosting both morale and efficiency.

It works. The enthusiasm for automation will eventually have to survive its messy encounter with software development's most chaotic parts, and that's where things like complex debugging, final human judgment, and ultimate operational accountability still rest firmly with people. But building a highly capable agent is its own challenging kind of coding.