When the Biggest Businesses Run on the Smallest Prompts

Jul 07, 2026

9 min read

2 views

42%

The surprising question hiding inside a booming medical specialty

What do a gastrointestinal procedure and an AI agent have in common? At first glance, almost nothing. One is a clinical act performed in a surgical center. The other is a software system that can be described in plain language and built to work while you do something else. Yet the deepest link between them is not technical. It is organizational.

Both reveal a shift in what creates value when work becomes repeatable, high volume, and operationally dense. In one case, gastrointestinal procedures now account for 28.3% of all Medicare ASC volume, which means an entire outpatient ecosystem has increasingly been shaped around one dominant service line. In the other case, the promise of agentic software is simple: describe the work, and the system builds the agent. That sounds like convenience, but it is actually a clue about the future of work itself.

The real question is not whether healthcare and AI belong in the same conversation. It is this: what happens when the highest value in a system comes from turning messy human effort into an industrialized, repeatable workflow?

That question matters because the same forces that made GI the dominant ambulatory procedure category are now making agents the dominant metaphor for business automation. In both worlds, the winners are not necessarily the ones with the flashiest technology. They are the ones who can convert complexity into a workflow that scales.

The hidden common denominator: volume rewards simplification

High volume changes everything. Once a procedure, process, or decision repeats enough times, the economics stop favoring bespoke effort and start favoring system design. In ambulatory care, this means facilities, staffing, scheduling, coding, and patient flow get optimized around a predictable stream of cases. In software, it means the valuable unit is no longer a one off task completed by a person, but a durable system that can execute a task repeatedly with less supervision.

This is why the GI boom is more than a specialty trend. It is a case study in what happens when a service line becomes so dominant that the surrounding organization reorganizes itself around it. The center of gravity shifts. The workflow becomes the product. The procedure is still important, but the real economic engine becomes the infrastructure that makes the procedure fast, reliable, and profitable.

Agentic software works the same way. The point is not merely to automate a single step. The point is to encode an operational pattern. When you say, “Describe the work,” you are not asking for a tool. You are trying to externalize a repeatable process into something that can run with minimal friction.

That is the deeper parallel: both GI volume and AI agents reward the translation of expert labor into repeatable architecture.

When a task repeats often enough, the competitive advantage stops living in the task itself and starts living in the system around it.

This is a useful mental model for any industry. In the early phase, success comes from skill. In the scale phase, success comes from standardization. In the mature phase, success comes from making standardization feel effortless.

The paradox of automation: the more human the work, the more it needs structure

A common mistake is to think automation only matters for trivial work. In reality, automation becomes most valuable when the work is important, repetitive, and full of coordination overhead. That is why outpatient GI is such a revealing example. It sits at the intersection of clinical expertise, patient logistics, reimbursement, throughput, and scheduling precision. The procedure itself may be specialized, but the business around it depends on operational discipline.

That is also why agentic systems are compelling. The phrase “just describe what you need” sounds casual, but behind it is a serious promise: highly contextual work can be decomposed into instructions, constraints, and triggers that software can execute. The more a business relies on recurring coordination, the more attractive it becomes to turn human effort into an agentic workflow.

Think about a common office process, like onboarding a new customer. There is often a chain of small actions: gather documents, send welcome emails, update a CRM, assign tasks, check compliance, follow up on missing items. None of those steps is individually glamorous. But together they create the experience. This is exactly the kind of work that agents are meant to absorb.

Now imagine the same logic in a surgical center. A procedure day is not just the procedure. It is intake, prep, room turnover, equipment availability, staffing, billing, and follow up. The dominant service line is never merely a medical category. It is an operational choreography.

That is the paradox: the more specialized the work, the more important the surrounding system becomes. Specialization creates volume. Volume creates repetition. Repetition creates the need for abstraction. And abstraction is where agents and industrial workflow design begin to matter.

Why the future belongs to workflow architects, not just operators

In every sector, there are three layers of value creation.

Doers execute individual tasks.
Operators coordinate tasks into reliable processes.
Workflow architects redesign the process so it can scale, adapt, and self correct.

The GI boom suggests that the organizations winning in high volume care are not just those with more procedures. They are the ones that understand how to turn procedures into throughput without losing quality. The agent builder message points to the same hierarchy in software: the value is in designing the process, not manually performing every step.

This shift matters because it changes how leaders should think about talent. In a workflow saturated environment, the rarest people are not necessarily the fastest executors. They are the people who can see the hidden sequence inside a messy process and turn it into a system. That may be a nurse manager optimizing patient flow, an administrator redesigning scheduling rules, or an operations lead defining when an agent should escalate versus proceed autonomously.

Here is a practical way to see it:

A task is what one person does once.
A workflow is what many people or systems do repeatedly.
A service line is what the organization becomes when a workflow is repeated enough times to dominate its economics.

This framework connects the two source ideas directly. GI is a service line that has become structurally central. Agents are a tool for making workflows behave more like service lines, meaning more repeatable, more measurable, and more scalable.

That is why the phrase “build the agent” is more than a product slogan. It captures a broader business truth: the modern organization is increasingly a set of callable workflows.

The real competitive moat is not automation, it is judgment

It would be easy to conclude that the future belongs to whoever automates the most. That is not quite right. Automation removes friction, but it does not remove judgment. In fact, as systems become more automated, judgment becomes more valuable because it determines where to automate, where to keep humans in the loop, and how to define success.

In a high volume medical setting, this is obvious. Standardization improves consistency, but there are always edge cases, complications, and exceptions. The best operations are not the ones that pretend exceptions do not exist. They are the ones that design for exceptions without letting them dominate the system.

Agentic software faces the same problem. A good agent cannot simply do everything. It needs boundaries. It needs escalation paths. It needs a sense of confidence thresholds. It needs rules for when to stop, ask, or defer. In other words, the most powerful automation systems are not the most autonomous. They are the ones with the best decision architecture.

This is where the two ideas become genuinely illuminating together. A dominant service line and an agent builder both expose the same strategic reality: scalable systems are not just machines for producing output. They are machines for preserving judgment at scale.

Scale is never just about doing more. It is about deciding what must stay human, what can be encoded, and what should be delegated to a system.

That insight matters in healthcare, but it also applies to any business feeling pressure to do more with less. The temptation is to automate the visible labor. The better move is to automate the coordination layer and protect the human layer where nuance matters most.

A simple framework for thinking about agentic businesses

If you want a mental model that links these ideas across industries, use this three part lens.

1. Identify the volume center

Ask: what work repeats enough to shape the business? In a healthcare setting, it may be GI procedures, imaging, intake, or scheduling. In a business setting, it may be invoicing, lead routing, support triage, or onboarding.

The volume center is where standardization creates the biggest payoff.

2. Map the coordination tax

Ask: what portion of the work is not the core work itself, but the handoffs around it? This is where time disappears. It is also where agents usually create the first meaningful ROI.

Examples include:

chasing missing forms
updating systems after a completed step
routing exceptions to the right person
generating reminders and follow ups
checking whether prerequisites are met

3. Encode the rules, not just the tasks

A strong workflow system does not merely automate clicks. It captures decision logic: when to proceed, when to pause, when to escalate, and when to retry. That is the difference between a brittle script and a durable agent.

This framework works because it moves the conversation away from “Can we automate this?” and toward “What part of this workflow deserves a machine, and what part deserves human attention?”

That is the crucial shift. The best organizations are not trying to eliminate humans. They are trying to make human effort more concentrated, more meaningful, and less absorbed by administrative drag.

Key Takeaways

Look for volume centers first. The most strategic opportunities are usually where repetition is already high and growing.
Automate the coordination tax, not just the core task. The biggest gains often come from handoffs, follow ups, and status management.
Treat workflows as products. If a process repeats often, it should be designed, measured, and improved like a product line.
Preserve judgment at the edges. The best systems know where humans are still better, especially in exception handling and escalation.
Think in architectures, not actions. A single automation is useful, but an encoded workflow is transformative.

Conclusion: the future is built by making repetition intelligent

The GI boom and the rise of agent builders are not random developments in unrelated industries. They point to the same structural truth: the future belongs to organizations that can turn repeated work into intelligent systems.

That does not mean everything becomes automated, sterile, or impersonal. It means the highest performing institutions will be those that can separate the parts of work that should be standardized from the parts that require human discretion. In other words, the winning move is not to replace expertise. It is to package expertise into systems that can repeat without degrading.

That is why a dominant service line and a simple prompt interface are more connected than they first appear. Both are signs that the next era of advantage will come from people who can see the workflow hidden inside the work. And once you see that, the question changes forever. You stop asking only what a business does, and start asking what machine it has quietly become.