If you're in software, you've likely heard some version of this declaration over the years: "We're just digitizing the filing cabinet." It's a fair metaphor — and one that defined the industry for decades. From Salesforce replacing rolodexes to SAP redefining inventory management, software has served as an organizational layer. It stored things. Indexed them. Made search faster. But it rarely did the work itself.
Today, that's changing.
We're watching the early days of a transition where software doesn't just manage labor — it becomes the labor. AI has made it possible for software to shift from being a passive manager into an active doer.
The implications are big. Software is no longer confined to the $300 billion SaaS market of seat licenses and cloud subscriptions. It’s stepping into a much larger arena: the $13 trillion U.S. labor market. It’s not just about reducing keystrokes for support agents — it’s about removing the agent entirely.
That might sound dramatic, but it’s already happening. There are AI systems answering tier-one support email, calling customers to collect payments, routing freight in logistics companies, conducting basic legal reviews, onboarding new hires across multiple time zones — all with minimal human oversight. They don't wait for the next shift to start. They don't call in sick. They don’t need to be reskilled.
Software now acts.
This is a different kind of disruption from previous waves of automation. The spreadsheet replaced the ledger book. Email replaced the fax. Each leap improved productivity, but still depended on human input to function. The current leap removes that dependency in meaningful ways.
As a team that’s been building software for a long time, we’ve felt the gravitational pull of this shift. In the past, choices like pricing models and user interfaces were centered around seat-based SaaS. You sold by the user — the more people using your tool, the more you made. This model worked well when software supported a human doing a job, but breaks down when software starts doing the job.
You can see it with companies like Zendesk. As of a few years ago, a large enterprise might spend $1.4M annually on the software, but over $75M on the people interacting with it. The opportunity isn’t in squeezing more out of the $1.4M — it’s replacing the $75M. And if the new AI system can deliver superior results, from first-response time to multilingual coverage, even charging $10M would feel like a bargain.
That starts to open doors into markets the software world largely ignored. Fields like optometry, nursing, logistics, collections — industries where margins are tight, labor costs are high, and software spend has traditionally been small. You'd never sell nine seats of Salesforce to a front desk staff in a small clinic. But an AI receptionist who works weekends, speaks Spanish, and syncs directly with insurance systems? That suddenly makes financial sense.
And as software gets better, the definition of a “niche” market changes. AI enables scale not by volume of users, but by number of tasks completed. Picking up the phone to settle overdue invoices in Farsi is a capability, not a feature — and it used to require hiring a specialist. Now it might be a prompt away.
We’ve also started seeing this change ripple outward. AI doesn’t just reshape software companies. It makes some previously unbuildable businesses viable. Imagine a startup that tried to create on-demand, door-to-door bike rentals in rural towns. Pre-AI, support and logistics would eat every margin. With AI operating the customer support, scheduling, and inventory coordination, suddenly the numbers start to work. AI lowers costs — not just of operations, but of experimentation.
There are deeper questions too. What does trust look like when a customer never interacts with a human? If outcome-based pricing becomes mainstream, how do B2B contracts evolve? In some ways, these are human questions, even as the workers become software.
One thing we try to stay conscious of: the goal isn’t just to reduce costs. It’s to unlock leverage. AI is a tool that tolerates frustration — whether it’s repetitive forms or regulatory snarls — in ways many of us can't or won’t. In doing so, it lets humans focus on where we're needed most, and reduces the friction of trying new ideas.
The phrase we come back to here isn’t about replacing jobs or minimizing headcount. That may happen in places. But the bigger vision is more creative: turning labor into code so that we can redistribute energy — both human and capital — toward what's next.
Systems of record helped businesses remember. Systems of engagement helped them communicate. Now, systems of action let them move.
Software, in other words, is becoming the worker. And for the first time, we’re starting to build with that in mind.
Today, that's changing.
We're watching the early days of a transition where software doesn't just manage labor — it becomes the labor. AI has made it possible for software to shift from being a passive manager into an active doer.
The implications are big. Software is no longer confined to the $300 billion SaaS market of seat licenses and cloud subscriptions. It’s stepping into a much larger arena: the $13 trillion U.S. labor market. It’s not just about reducing keystrokes for support agents — it’s about removing the agent entirely.
That might sound dramatic, but it’s already happening. There are AI systems answering tier-one support email, calling customers to collect payments, routing freight in logistics companies, conducting basic legal reviews, onboarding new hires across multiple time zones — all with minimal human oversight. They don't wait for the next shift to start. They don't call in sick. They don’t need to be reskilled.
Software now acts.
This is a different kind of disruption from previous waves of automation. The spreadsheet replaced the ledger book. Email replaced the fax. Each leap improved productivity, but still depended on human input to function. The current leap removes that dependency in meaningful ways.
As a team that’s been building software for a long time, we’ve felt the gravitational pull of this shift. In the past, choices like pricing models and user interfaces were centered around seat-based SaaS. You sold by the user — the more people using your tool, the more you made. This model worked well when software supported a human doing a job, but breaks down when software starts doing the job.
You can see it with companies like Zendesk. As of a few years ago, a large enterprise might spend $1.4M annually on the software, but over $75M on the people interacting with it. The opportunity isn’t in squeezing more out of the $1.4M — it’s replacing the $75M. And if the new AI system can deliver superior results, from first-response time to multilingual coverage, even charging $10M would feel like a bargain.
That starts to open doors into markets the software world largely ignored. Fields like optometry, nursing, logistics, collections — industries where margins are tight, labor costs are high, and software spend has traditionally been small. You'd never sell nine seats of Salesforce to a front desk staff in a small clinic. But an AI receptionist who works weekends, speaks Spanish, and syncs directly with insurance systems? That suddenly makes financial sense.
And as software gets better, the definition of a “niche” market changes. AI enables scale not by volume of users, but by number of tasks completed. Picking up the phone to settle overdue invoices in Farsi is a capability, not a feature — and it used to require hiring a specialist. Now it might be a prompt away.
We’ve also started seeing this change ripple outward. AI doesn’t just reshape software companies. It makes some previously unbuildable businesses viable. Imagine a startup that tried to create on-demand, door-to-door bike rentals in rural towns. Pre-AI, support and logistics would eat every margin. With AI operating the customer support, scheduling, and inventory coordination, suddenly the numbers start to work. AI lowers costs — not just of operations, but of experimentation.
There are deeper questions too. What does trust look like when a customer never interacts with a human? If outcome-based pricing becomes mainstream, how do B2B contracts evolve? In some ways, these are human questions, even as the workers become software.
One thing we try to stay conscious of: the goal isn’t just to reduce costs. It’s to unlock leverage. AI is a tool that tolerates frustration — whether it’s repetitive forms or regulatory snarls — in ways many of us can't or won’t. In doing so, it lets humans focus on where we're needed most, and reduces the friction of trying new ideas.
The phrase we come back to here isn’t about replacing jobs or minimizing headcount. That may happen in places. But the bigger vision is more creative: turning labor into code so that we can redistribute energy — both human and capital — toward what's next.
Systems of record helped businesses remember. Systems of engagement helped them communicate. Now, systems of action let them move.
Software, in other words, is becoming the worker. And for the first time, we’re starting to build with that in mind.