Wayne Liu is the Chief Growth Officer and President of Perfect Corp America.
For decades, software has been a tool, a silent partner that helps humans record, calculate and manage the complexity of business. “Systems of record” can track what people do, and “systems of engagement” make it easier for them to do it.
But something profound has changed. Artificial intelligence empowers software to act, not just assist. From my observations, many programs are no longer passive databases waiting for human commands; they’re becoming active agents for reason, decision making and execution. Arguably, software is, quite literally, starting to work for itself.
This moment marks the birth of what I and others call the “Agentic Economy”: A new chapter in which code is not merely a tool of labor, but labor itself. The implications reach far beyond automation. They stand to redefine how value is created, how capital is deployed and how organizations measure productivity.
This article builds on ideas I first introduced in my earlier piece, expanding the conversation from AI-enabled business models to AI-driven labor itself.
Capital Is Becoming Labor
Once software begins to act, it stops being a tool and starts becoming a worker. That changes everything about how we value it.
In my view, this new generation of AI changes the basic economic equation of technology. For decades, capital built tools; the labor force used them. Now, capital itself can perform labor through computation and code. Every investment in GPUs or model training can directly create productive capacity. I believe that the relationship between software spending and labor spending is converging into a single continuum.
Think about it this way: The U.S. software-as-a-service (SaaS) market was worth around $140.7 billion in 2024. By contrast, according to the Bureau of Labor Statistics (BLS), based on “reports submitted to state workforce agencies by every employer covered by UI or by UCFE” in the United States, annual wages totaled $11.7 trillion in 2024. When software starts performing work rather than assisting, the total addressable market for software vastly expands. I expect that in the future, software companies will no longer be selling tools—they’ll be competing for labor budgets.
From Selling Seats To Selling Productivity
Selling “seats” is a traditional SaaS pricing model; the more users a company has, the more it pays for SaaS software. That model makes sense when the output is scaled relative to headcount.
But what happens when an AI agent performs the work of 10 people? It breaks the SaaS pricing model of charging per user, because productivity is no longer solely defined by human access, but also by computational performance.
I see this as a fundamental structural reset. I predict that the next generation of software companies will have to price by results, not logins. If you’re a SaaS leader, I recommend that you audit your SaaS stack today: Identify one workflow where agentic automation could replace seats with outcomes.
The Principles Of The System Of Labor
Here’s how I believe the path toward “systems of labor” will look:
1. Commoditization: According to research, “technological advancement can often lead to commoditization.” In my view, the cost of reasoning is approaching zero, with access to knowledge and analysis becoming as ubiquitous as electricity, democratizing expertise across industries.
2. Standardization: As explained in a research paper, “standardization is an essential prerequisite for the implementation of new technologies.” Automation begins where outcomes are easy to standardize or measure. Tasks that can be clearly standardized, such as coding, accounting and logistics, will likely be transformed first.
3. Asymmetry: In economics, Investopedia notes, asymmetric information refers to the “uneven amount or quality of material knowledge held by those engaged in a transaction.” I expect that AI mastery will advance unevenly, with digital and data-rich domains evolving the fastest and physical and ambiguous work lagging behind.
By understanding commoditization, standardization and asymmetry, leaders can predict which industries will be reshaped first and in which industries human creativity will remain indispensable.
Strategic Implications: Management In The Agentic Era
As software assumes cognitive tasks, leadership must evolve as well. I expect that, moving forward, many managers will orchestrate networks of agents, akin to managing teams of people. This will require a new skill set rooted in data fluency, prompt engineering and comfort with stochastic systems that learn dynamically.
Instead of assigning tasks, managers should define outcomes, monitor feedback loops and optimize the interactions between humans and agents. The organizations that stand to become the most successful will treat AI as an integrated part of their operating rhythm.
Why This Change Matters
Beyond cost reduction, AI agents can rapidly scale to handle spikes in demand, sustain performance in repetitive roles and provide consistent service across languages and regions. They can help maintain compliance, audit their own reasoning and adapt to new rules overnight.
More importantly, they can unlock entirely new markets. Consider small businesses that spend little on software. With agentic software, businesses can “hire” digital labor instead of buying software, expanding productivity to areas beyond what’s possible or feasible in the traditional labor market. AI labor can bring expertise and capacity to the long tail of the software economy.
The New Industrial Shift
I see data centers turning into AI factories, large language models becoming the new machinery and AI agents as the new workers. The raw materials? Data and context. The outputs? Insights, designs and decisions.
This transformation mirrors the industrial revolution’s impact on physical work, but now it’s cognition that is being industrialized. The question shouldn’t be whether AI will replace humans; it should be how human and AI workforces will integrate. Will humans set goals and apply judgment, leaving agents to handle execution at a massive scale? The answer isn’t yet clear.
Redefining Software
In the decades since Tom Kilburn wrote a program with only 17 instructions to run on the Manchester “Baby” computer, software has recorded what humans do and enabled them to complete tasks. Now, software is starting to do the work itself.
In my view, the companies that thrive in the next decade will measure success not by the number of users they serve, but by the productivity of their systems. Pricing, management and strategy will stand to all revolve around one core metric: outcomes delivered.
As I see it, the next great software companies won’t be built on better databases or UI/UX. They’ll be built on networks of AI agents that think, act and deliver value autonomously.
And that, to me, marks the true beginning of the Agentic Economy.
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?
