AI can now not only help with the work but increasingly do the job entirely across many service industries. 

We have been in a time of technology enablement and copilots. In healthcare, dozens of large and thriving tech-enabled platforms like Headway, Alma, or Omada built businesses by making it easier for humans to deliver, document, process, and manage care. AI-powered tools like Abridge help clinicians with much needed documentation. But now, companies like Doctronic are providing care directly through AI. They're not only the assistants, but the sole practitioners.

In law, tools like CaseText and Harvey help lawyers research and draft. But we're also beginning to see AI-powered legal agents that do the work, not just make it easier. Soon, agents may themselves hold attorney client privilege or be able to execute a financial trade.

It is becoming easier and easier to “sell the work” not the service, a concept Sarah Tavel described in a couple of A+ blog posts. But now, we can push it a step forward by selling the work directly to the end user–the patient, plaintiff, customer, etc–instead of the business providing the service. We are finding ourselves at USV most interested in products and platforms that are doing this direct motion.  

With this change, previously held definitions are dissolving. The line between “tool” and “worker” is disappearing. Markets known for enterprise sales and pricing are becoming increasingly consumer oriented. And the concept of “billable hours” is quickly becoming antiquated.

Services have generally been measured in time. Lawyers charge by the hour. Therapists see clients in 50-minute increments. Consultants submit timesheets. A tutor bills the student for the length of each session, colleges charge by semester. In this world, technology was a supporting actor with increasingly powerful tools that help humans do the work faster, better, or more affordably.

With AI moving from copilot to pilot, we need new ways of defining and pricing value. When a machine can not only produce a legal brief or generate therapy notes instantly, but become the lawyer, therapist, physician, advisor, accountant, architect, teacher itself, the amount of time spent becomes irrelevant. What matters is quality of product not time spent producing. In situations where software is talking to software, usage based billing has emerged as a new king. But in this ecosystem, we’ll also need new business models to evolve and one possibility is that they will be ones that value billable outcomes. 

This opens the door to a long-needed realignment. Many industries have suffered from the perverse incentive that more time equals more money. Now we can shift the equation from how much we do to how well we do it. But with it, we will need to rethink and reinvent the model of what these outcomes are worth. Is an outcome worth the sum of imagined time previously spent, some equation of usage based or aggregated compute, or, ideally, something new entirely? Can we readjust cost centers away from where they have previously been, opening up new forms of access and re-designating the spend towards ways to push capabilities further? We are in a moment of extreme progress towards outcome alignment but also an opportunity to redefine how to value it. 

AI can now not only help with the work but increasingly do the job entirely across many service industries. 

We have been in a time of technology enablement and copilots. In healthcare, dozens of large and thriving tech-enabled platforms like Headway, Alma, or Omada built businesses by making it easier for humans to deliver, document, process, and manage care. AI-powered tools like Abridge help clinicians with much needed documentation. But now, companies like Doctronic are providing care directly through AI. They're not only the assistants, but the sole practitioners.

In law, tools like CaseText and Harvey help lawyers research and draft. But we're also beginning to see AI-powered legal agents that do the work, not just make it easier. Soon, agents may themselves hold attorney client privilege or be able to execute a financial trade.

It is becoming easier and easier to “sell the work” not the service, a concept Sarah Tavel described in a couple of A+ blog posts. But now, we can push it a step forward by selling the work directly to the end user–the patient, plaintiff, customer, etc–instead of the business providing the service. We are finding ourselves at USV most interested in products and platforms that are doing this direct motion.  

With this change, previously held definitions are dissolving. The line between “tool” and “worker” is disappearing. Markets known for enterprise sales and pricing are becoming increasingly consumer oriented. And the concept of “billable hours” is quickly becoming antiquated.

Services have generally been measured in time. Lawyers charge by the hour. Therapists see clients in 50-minute increments. Consultants submit timesheets. A tutor bills the student for the length of each session, colleges charge by semester. In this world, technology was a supporting actor with increasingly powerful tools that help humans do the work faster, better, or more affordably.

With AI moving from copilot to pilot, we need new ways of defining and pricing value. When a machine can not only produce a legal brief or generate therapy notes instantly, but become the lawyer, therapist, physician, advisor, accountant, architect, teacher itself, the amount of time spent becomes irrelevant. What matters is quality of product not time spent producing. In situations where software is talking to software, usage based billing has emerged as a new king. But in this ecosystem, we’ll also need new business models to evolve and one possibility is that they will be ones that value billable outcomes. 

This opens the door to a long-needed realignment. Many industries have suffered from the perverse incentive that more time equals more money. Now we can shift the equation from how much we do to how well we do it. But with it, we will need to rethink and reinvent the model of what these outcomes are worth. Is an outcome worth the sum of imagined time previously spent, some equation of usage based or aggregated compute, or, ideally, something new entirely? Can we readjust cost centers away from where they have previously been, opening up new forms of access and re-designating the spend towards ways to push capabilities further? We are in a moment of extreme progress towards outcome alignment but also an opportunity to redefine how to value it.