A second robot arrives and the business case gets harder

Every device demo I have sat through ends on the same slide. One large throughput number, a photograph of a treatment room lit better than any room I have ever worked in, and a sales lead waiting for me to be impressed. The question I ask straight after that slide is deliberately boring. How many days a year will this thing actually be switched on?

That question got more interesting in July. At the 14th World Congress for Hair Research, Seoul-based Puncture Robotic showed HAIRO, a robotic FUE platform. Coverage from HairTransplantClinics.com describes 20-megapixel imaging, machine-learning extraction of up to 2,400 follicles an hour, pre-operative simulation, and an algorithm that skips single-hair units so multi-hair grafts get priority. It carries Class III certification from China. It has no FDA clearance and no CE mark.

For about a decade, robotic FUE has meant one machine. A single-vendor market is easy to evaluate, because the decision is binary and the price is whatever the vendor says it is. Two credible platforms is a more uncomfortable position. Prices move, claims get compared, and somebody senior reads a trade article and asks why you have not bought one yet.

So here is the method I use. It has very little to do with the robot.

Find the constraint before you price anything

Extraction speed only matters if extraction is where the queue forms. In the networks I have worked in, it usually is not. The limiter has been surgeon hours, trained hands available on the day, consultation-to-surgery conversion, or how many rooms the lease gave us.

The ISHRS practice census puts member surgeons at around fifteen surgeries a month. That is not a machine utilisation problem. That is a diary and demand problem, and no amount of imaging resolution fixes it.

If you cannot name your binding constraint in one sentence before the demo starts, you will end up buying the machine that relieves whichever constraint the vendor sells against. I have watched that happen. The equipment performs exactly as promised and the monthly case count does not move, because the real bottleneck was three steps downstream the whole time.

The denominator nobody puts on the slide

Vendors quote cost per graft at planned volume. Build the same number at sixty percent of planned volume and see whether the case survives. If it only works with a full diary, it does not work, because the diary will not be full in month two of a new market and it will not be full the week your lead surgeon takes leave.

Then add the parts that never reach the slide. Service contract. Consumables per graft. Downtime, and how the contract defines downtime. Engineer response time, which is a very different number in a mature market than in a growing one. In one market I worked in, the nearest field engineer for a far simpler piece of imaging equipment was a flight and a customs form away, which turned a two-day fault into a two-week hole in the schedule.

Capex is one conversation. The service reality is a five-year one.

Automation moves the skill, it does not remove it

Robotic extraction does not touch site creation, graft handling under scopes, or placement. Those steps stay human and they absorb the labour the robot frees up. Anyone selling you a headcount reduction is selling you a headcount reshuffle.

Which makes each market’s delegation rules a first-order input rather than a footnote. Turkey approved a standard in February 2026 creating a certified assistant-practitioner role for hair transplantation, with 104 hours of training and physician supervision still mandatory, as Medical Tourism Watch reports. India is moving the other way. Finshots reports the government pushing to classify transplantation strictly as surgery, performed only by registered medical practitioners in licensed facilities.

Same machine. Two very different labour models around it, and therefore two very different returns. A platform that lets certified assistants run long extraction sessions under supervision earns its keep in the first market and sits waiting for scarce physician time in the second.

Clearance is a policy problem, not a purchase problem

HAIRO’s certification position is the part operators should sit with properly. Cleared in one large market, not yet cleared in the two that most brand standards are written against.

If your operating standard names a device, you have just created two brands under one logo. The clinic with the machine follows the document. The clinic without it improvises, and improvisation is exactly what an audit programme is supposed to catch.

Define the graft survival expectation, the acceptable transection band, session duration limits, documentation requirements, and the competency sign-off for each step. A device then becomes one of several compliant ways to meet the standard, and a franchisee in a market without clearance is not automatically non-compliant. This version is slower to write. It is the only one that survives contact with a regulatory map that changes country by country.

Write the standard at the level of the outcome, not the level of the equipment. Devices get cleared, withdrawn and superseded. Transection rate does not.

Assume everyone will be able to rent it

The pattern is already obvious on the software side. Zenoti announced eleven purpose-built AI agents covering medspa workflows this spring, from reception and lead handling through charting, inventory and retention. Epic spent its August users meeting folding agents and ambient charting into the platform itself. Capability that was a differentiator two years ago is now a subscription line, available to your competitor at the same price it is available to you.

Hardware follows the same arc, more slowly. Mordor Intelligence sizes the transplant robot segment near $800 million inside a hair restoration services market heading toward $12.5 billion by 2031. That gap is the point. The machine is a component, not the business.

So the useful question about any automation is what you will still own once your competitor has the same thing. Usually the answer is candidacy screening discipline, planning quality, the audit trail, and what your network does on the day a case goes badly.

What I would actually do

Buy nothing for six months. Spend that time collecting the baseline you will need to judge the machine: transection rate by operator, session duration against graft count, revision and repair rates, and how many surgical days each site genuinely runs. Most networks cannot produce those numbers on request, which means they cannot evaluate a robot against anything except the vendor’s own slide.

Then, if the case still stands, put it in one site with one surgeon and one clearly defined comparison, and let it run long enough to hit a bad month.

The honest reason to say no to a robot is rarely the price. It is that you were never measuring the thing it claims to improve. The machine stays outside your control until the measurement is inside it. Start where your control already is.