The most common invasive medical procedure in the world is still done by hand. Every year, tens of millions of blood draws rely on a trained eye, a steady wrist, and a patient who holds still. In June, a machine did it alone in Denmark for the first time.
In a peer-reviewed multicenter study published in Clinical Chemistry, the robot hit a 94.5% first-attempt success rate across 1,633 patients, with a 0.6% hemolysis rate and a median procedure time of about one minute and 49 seconds.
Vitestro just closed a $70M Series B backed by Mayo Clinic, Labcorp, and Sutter Health, and is pursuing a US FDA De Novo pathway, while DARPA funds autonomous vascular access for field transfusion.
The case for the machine
Start with the numbers nobody disputes. Across three outpatient phlebotomy departments in the Netherlands, Aletta succeeded on the first attempt 94.5% of the time when it located a suitable vein. The rate held at 97.4% in patients with a body mass index above 30, and 93.4% in patients older than 65.
In patients who reported difficult venous access, it still landed 92.7%. The earlier pivotal study of 609 participants reported a 95% first-stick rate, a hemolysis rate of 0.6%, and no serious adverse events. Hemolysis is the quiet killer in diagnostics: blood cells burst during collection, and the sample has to be redrawn.
The economic argument is just as direct.
Phlebotomy is the last fully manual step in the laboratory pipeline. Staffing is the constraint, not demand. Northwestern Memorial's medical director of pathology, Gregory Retzinger, told CAP Today about a winter day when phlebotomists called off work and the outpatient testing center backed up. "We couldn't collect samples from our outpatients," he said. "We can't keep going through this."
He reached out to the company for the trial. Not because the robot out-performs his own 92% staff rate. Because drawing blood reliably requires bodies, and bodies call in sick.
That is the real unit economics of automation: not 94.5% versus 92%, but guaranteed capacity when the morning lab is short staffed. One supervisor oversees three Aletta units at once.
Then add the frontier application. In May 2026, DARPA's Biological Technologies Office opened an SBIR topic for a Smart Whole Blood Field Transfusion system, a disposable, under-five-pound device that must "autonomously perform vascular access and IV placement." The agency names that subsystem as the technical crux. The same robotics, AI, and ultrasound stack that fills a tube in Utrecht is the piece DARPA needs for a $300 field transfusion device.
That is the bridge from outpatient labs to field medicine: the autonomy stack is no longer a lab novelty. It is infrastructure for remote care.
The case for the human
Hold the other side of the scale.
Trained US phlebotomists average a 93–97% first-stick rate. Aletta's 94.5%, measured only in patients where the machine found a suitable vein, starts to look less like a leap and more like parity. The qualifier matters: the denominator excludes everyone the system declines.
And the list of declines is long. The current application does not cover pediatric patients, patients under 16, or anyone with burns, mastectomies, scarring, arm-limited disabilities, or needle phobia. Patients needing hand draws, special positioning, or emergency intervention are out. So is anyone whose veins are hard enough to need a skilled assessment in the first place.
The machine automates the easy draws. The hard ones stay human. The company frames the deployment that way, with patients choosing the robot or the phlebotomist at high-volume sites.
The deeper objection is evidence. Every performance number published so far comes from its own trials. The European hospital partners issued enthusiastic statements about "very positive" patients, but published no independent success rates, hemolysis figures, or complication data. As of late 2025, the Center for Phlebotomy Education's review could find no independent operational validation anywhere.
And the regulatory gap is real. Aletta is CE-marked for Europe. It is not FDA-approved, and is not available for sale or use in the United States. Vitestro is pursuing a De Novo pathway, a first-of-kind review in which the FDA examines an autonomous invasive procedure rather than an assistive device. Medical Daily's August coverage points to the watchpoint: any device performing an invasive procedure autonomously will need authorization for that specific function.
This is a timing problem.
The data speaks before the brand does
Robot first-stick rate, 1,633 patients
Multicenter Dutch trial in Clinical Chemistry: 94.5% success when a suitable vein was found, 97.4% in BMI-over-30 patients. · Vitestro trial data via Medical Daily, 2026
| Metric | Aletta (robot) | Trained phlebotomist |
|---|---|---|
| First-stick success | ✔ 94.5% (suitable-vein sample) | ◐ 93–97% typical |
| Procedure time | ✔ ~1 min 49 s median | ◐ varies with patient |
| Patient coverage | ✗ Healthy adults 16+ only | ✔ All ages, all anatomies |
| US availability | ✗ Not FDA-approved | ✔ In use nationwide |
The difference between 94.5% and 92% will not decide the market. The difference between "a phlebotomist available" and "no sample collected" will.
What the next 24 months actually test
Three questions separate the clinical success from the commercial one.
First, the US readout. Mayo Clinic and Northwestern plan rigorous validation protocols before broad adoption, and that evidence set will drive the FDA De Novo decision.
Second, the revenue model. High-volume outpatient sites are the beachhead, but the machine must hold its 94.5% at deployment scale, after launch delays and in systems that run it alongside human staff, the same promised-versus-reality chasm every robotics roll-out crosses. As we wrote in July, humanoid robots reached European factory floors with exactly this gap: strong pilots, unproven fleet economics.
Third, the field-medicine bridge. DARPA's SWiFT request explicitly demands autonomous IV placement in a $300 disposable device. If that stack federates out of lab robotics, the same class of machine moves from outpatient centers toward disaster kits, ambulances, and remote care. That is where a $70M round becomes a platform story.
Watch the FDA pacing. Watch independent readouts. Watch whether the easy-draw segmentation holds.
A robot drew blood in Denmark in June. It was precise, quick, and reliably available.
The humans still drew the blood the robot could not.
