Secretaries Are a Prototype for Systems of Action
My friend Horace offered “There is a belief, not widely known, that the best legaltech ever created was secretaries.” I agree. Why?
I frequently tell a joke that is not a joke.
What is an attorney’s favorite application?
People usually bid Microsoft Word, Outlook, or a research tool. I respond with
“A secretary!”
The most valuable “application” in an attorney’s working life is not a tool they operate directly. It is a system that observes their work, infers intent, acts autonomously within understood bounds, learns preferences over time, and shields them from the operational complexity of other systems. In other words, a system of action.
Much of the current conversation about AI products focuses on capability. What can the system do, how broadly can it generalize, and how quickly is that frontier advancing. These questions matter, but they often obscure a more fundamental constraint in professional environments: human cognition.
Unallocated time, attention, and cognitive effort are increasingly scarce. Many AI products impose a high fixed cost before they deliver meaningful value. They require users to learn new mental models, new workflows, and new modes of interaction. Even when a tool is extremely powerful, the effort required to decide what to ask, how to ask it, and how to evaluate the output can outweigh the benefit.
For many people, a general purpose interface still feels like a command line. Capable of almost anything, but demanding fluency, experimentation, and sustained attention. That is not how most workdays are structured.
Most people’s days are dominated by very specific somethings. Deadlines. Drafts. Emails. Scheduling. Follow ups. Coordination across people and systems. The relevant question is not whether a general system can handle these tasks in principle. It is how people want those tasks to be handled, and how much thinking they are willing to spend orchestrating the handling.
This is where the secretary becomes a useful prototype rather than a nostalgic metaphor.
One reason this pattern is so consistently missed is experiential. Many people building productivity technology have never worked as a secretary or closely alongside a great one. They have not lived the work of observing, inferring, buffering, and optimizing on behalf of a principal. As a result, products are often designed around explicit tasks and interfaces rather than around the gradual accumulation of optimizations that reduce cognitive effort over time.
A great secretary does more than execute explicit instructions. They observe a principal’s work and begin to recognize patterns. When confidence is high, they act without being asked, taking a task as far as their authority and capability allow. They execute asynchronously and revert only when more input is required. Over time, they learn preferences that are rarely written down. Formatting quirks. Communication styles. Risk tolerances. These preferences become implicit knowledge that is costly to recreate and therefore highly sticky.
Secretaries also act as a shim between the principal and a long tail of tools and systems. Time and billing. Travel booking. Filing systems. Internal workflows. The principal does not need to master these systems because the cognitive and operational burden has been delegated. As a result, value accrues to the system of action, not to the underlying systems of record.
Seen this way, the secretary is not just support staff. It is an existence proof. A demonstration that the most durable productivity gains come from systems that conserve cognition, learn continuously, and earn the right to act on a user’s behalf.
If human cognition is the constraint, then the central design question for AI products is not how general they are, but how effectively they function as systems of action. That reframes the platform versus product debate and suggests that many of the things great secretaries do are worth our time, attention, and cognition. They are the prototype for experiences that make us more effective.
“Efficiency is concerned with doing things right. Effectiveness is doing the right things.”
-Peter Drucker