A 58-year-old man with a history of hypertension and a 30 pack-year smoking history presents to the emergency department with chest pain that started 2 hours ago. He describes the pain as a pressure-like sensation radiating to his left arm, associated with diaphoresis and nausea. His blood pressure is 148/92 mmHg, heart rate is 102 bpm, respiratory rate is 18/min, and oxygen saturation is 96% on room air. An ECG shows ST-segment elevation in leads II, III, and aVF. Troponin is pending.

Which of the following is the most appropriate next step in management?

A) Administer aspirin 325 mg and activate the cardiac catheterization lab
B) Obtain a chest X-ray and wait for troponin results
C) Start a heparin drip and admit to the general medicine floor
D) Perform an echocardiogram to assess wall motion abnormalities
E) Administer nitroglycerin and observe for 30 minutes

Questions like these are considered the gold standard in medical education. They are the backbone of the USMLE Step exams, shelf exams, and virtually every high-stakes assessment a medical student will face. However, when you go to a clinic, you are NOT given a neat descriptor of information. You are given a history, some vitals, and A PATIENT in a hospital bed.

There is a clear gap between the vignette question and the experience given in the hospital. The vignette makes assumptions. It assumes that you will manage to extract the pertinent information from the patient that will lead you to the diagnosis. This, however, can be easier said than done. The communication that occurs between a patient and a provider is an art of its own that needs to be mastered.

Many medical schools acknowledge this and have long tried to supplement this training through clinical skills exams called OSCEs, Objective Structured Clinical Examinations. In an OSCE, students are placed in front of standardized patients, trained actors portraying specific clinical scenarios, and are assessed on whether they can actually extract the information the vignette hands to them for free.

The problem is scale. A single OSCE session requires coordinating trained actors, faculty evaluators, physical exam rooms, and scheduling windows that work across an entire cohort. The result is that most medical students get two or three of these encounters before a high-stakes clinical skills exam. Some get fewer. Once rotations begin, the number drops to zero, replaced by real patients in real clinical environments that operate on completely different terms.

This is the initial pain point that Praxora aims to solve: finding a way to provide students with on-demand clinical skills practice, anytime, anywhere.

However, the overarching goal of Praxora is much greater. I personally find clinical vignettes boring. Instead, I want the process of learning medicine to be low-stakes, fun, and exciting. In other words, I want to gamify medicine.

Take the job of truck driving. There are many people who work this job legitimately as a means to earn a living. However, oddly enough, there is also a cohort of people who simulate the job of a truck driver, spending thousands of dollars and hours building elaborate simulation setups in their living rooms.

Nobody is making them do this. There is no exam at the end, no grade, no credential on the line. They do it because the simulation itself is compelling. The act of navigating a rig through a mountain pass at night, managing gear changes and weather and blind corners, is genuinely engaging when the stakes are low and the feedback is immediate.

That is what medicine should feel like to learn. Not a multiple choice question with five options and a two-line stem. A conversation. A person sitting across from you, telling you their chest has been hurting for two hours, and you having to figure out what questions to ask next.

When learning feels like play, people do more of it. They practice at 11pm not because they have to but because they want to see if they can crack the case. They run the scenario again after getting feedback because the feedback actually meant something. They share cases with their friends the way gamers share levels. This is the version of medical education I want to exist. Praxora is the first step toward it.

We are releasing a minigame version of our app in the next few weeks, so if you are interested in Praxora, please visit our website: praxora.org.

NO. You’re thinking in the wrong direction. I’ll spend the next 750 words explaining why.

From what I’ve noticed, there are two ways to start a business that actually works:

1. Spray and Pray

2. Search and Destroy

Subscribe now

Spray and Pray

Spray and Pray is exactly what it sounds like. You build things. A lot of things. You ship, you iterate, you ship again. Eventually, something hits.

The poster child for this is Peter Steinberger, better known online as steipete. He’s the creator of OpenClaw, an open-source AI agent that crossed 370,000+ GitHub stars in a matter of months.

If you scroll through his GitHub, the pattern is obvious. Lots of small projects. Most of them useful. Most of them nobody noticed.

Then he shipped Clawdbot, which became OpenClaw, and it took off. On February 14, 2026, Steinberger announced he was joining OpenAI to work on agents. Sam Altman publicly welcomed him on X, calling him "a genius with a lot of amazing ideas about the future of very smart agents."

That’s the spray and pray model in one career. You don’t know which seed becomes the redwood, so you plant a forest. You stay CHEAP, you stay FAST, and you treat every shipped product as a lottery ticket that also teaches you something. As Steinberger states in his philosophy:

"Ship beats perfect" - I build tools to solve my own problems, then share them with the world. Currently exploring how AI changes everything about software development.

The catch is that this method has a hidden tax. You need the SKILLS, the TIME, and the TOLERANCE for ten flops before one lands. For most people, and especially for people with a day job, a clinical rotation, or a family, that tax is too high.

Search and Destroy

The opposite move is to find one specific problem, CONFIRM that real people have it, and then build ONLY the thing that solves it. NOTHING ELSE.

This is what Rob Fitzpatrick's The Mom Test is about. It’s a short book about how to talk to potential customers without being lied to. The framing is that everyone will lie to you about how good your business idea is, because they want to be polite. So you can't ask people if your idea is good. You have to ask about their actual life and their actual past behavior. Talk about them, not about you. The signal isn't "I would buy that." The signal is "I already tried to solve this, and here's the duct-taped mess I'm using right now."

That’s the searching part. The destroying part comes after. Once you’ve found a real, bleeding, recurring problem, you build the cleanest solution to it. NOTHING ELSE. NO BULK. NO FAT. ONLY LEAN, MEAN RESULTS.

Steve Jobs said the same thing at WWDC 1997:

“You’ve got to start with the customer experience and work backwards to the technology. You can’t start with the technology and try to figure out where you’re going to try to sell it. And I’ve made this mistake probably more than anybody else in this room, and I’ve got the scar tissue to prove it.”

The customer experience is the seed of the company. The technology, the brand, the domain name, the camera, those are downstream.

If you don't have the seed, the rest is decoration.

Conclusion

Spray and pray works if you can ship cheaply and constantly. Search and destroy works if you have access to a real customer with a real problem.

This is why clinicians are the BEST health tech founders of our era.

A clinician sits at the workflow. They watch an ER doctor click roughly 4,000 times in a single shift. They watch one hospital take 62 clicks to order Tylenol. They watch the diabetic patient who can’t read the discharge instructions.

Hence, clinicians are not guessing what the customer wants. They tend and care for the TRUE CUSTOMER everyday: the patients.

Hence, if you are looking to search and destroy, I would recommend by reading The Mom Test and going out to ask potential customers what their pain points are.

Looking at this graph, you might assume the United States has the world’s best healthcare system. After all, the U.S. spends nearly twice as much per citizen as the average country in the Organisation for Economic Co-operation and Development (OECD). However, the reality is starkly different:

The United States doesn’t just fall short of being the best, it actually performs below average, and this is not just in life expectancy, but other metrics as well:

Plotted together, the U.S. is a clear outlier among high-income countries: paying more, getting less.

The cost of our inefficient system is that a great share of Americans skip care entirely because they can't afford it:

As you look at these figures, consider a time when you or a loved one deferred care because of the potential costs, only to face a much steeper physical and financial cost down the road. This issue is felt among the majority of us, regardless of race or ethnicity. This clip below illustrates that reality perfectly: a Trump voter explains why his encounter with affordable care in Japan radicalized his perspective.

The following video will only play for one minute.

I used to hope policy would fix this. I don't anymore, at least not within a timeframe that matters for the patients I'll see in residency. So I've started looking at the other lever this country actually pulls well: technology.

Here's where most clinicians stop reading, because "tech will fix healthcare" is what every founder with a pitch deck has been saying for years, but the cost curve hasn’t bent. I think I know why.

Health tech has been built by people who don't see patients, with clinicians brought in as advisors rather than founders. That arrangement lets companies check the "clinical input" box while still optimizing software for the people writing the checks: administrators, payers, investors. Clinicians don't need a louder seat at someone else's table. We need to build the table.

We are entering an era where clinical expertise matters more than coding proficiency. With AI handling the heavy lifting of initial development, frontline providers are now empowered to evolve from problem-spotters to prototype-builders and company founders.

Over the next four years I'll be building, writing, and documenting what it actually takes for a clinician to become a founder rather than a consultant: what works, what breaks, and what the people already doing this wish they'd known earlier. If I'm wrong, you'll watch me be wrong. If I'm right, you'll have a roadmap.

New post, biweekly on Wednesday mornings. Under 800 words. No filler.

If any of this resonates, subscribe and push back in the comments. I learn faster when you do.

Subscribe now

Thank you for your time.

Best regards,
Pouya