Quiet Breakout: The First Baby Born From a Robot Was Conceived 3,700 Kilometres Away

The healthy boy was delivered at 38 weeks, weighed 3.3 kilograms, and scored a 9 on the Apgar test. By the metrics the obstetrician cares about, an unremarkable birth. The unusual part happened ten months earlier, in a small lab in Guadalajara, Mexico, when an embryologist set a single oocyte under an inverted Olympus microscope and stepped back. From that point on, every other action was performed either by software or by a remote operator in Hudson, New York, about 3,700 kilometres up the continent. A non-contact laser immobilised a sperm cell mid-tail. A robot pipettor positioned it in a hollow needle, computer vision steered the needle through the oocyte’s outer shell, and an AI confirmed the needle had cleared the egg before the robot withdrew. Twenty-three steps, none touched by a human hand at the bench. The case report, published in Reproductive BioMedicine Online in 2025, describes the first child conceived by remote, semi-autonomous intracytoplasmic sperm injection.

That was prologue. Six months later the same group, reporting in Human Reproduction, described five live births from a fully integrated automated IVF lab in which oocyte retrieval, denudation, sperm preparation, and ICSI were all performed by robots under AI control. The fertilization rate was 64.3 percent against 81 percent for a sibling-oocyte manual control arm. The blastocyst rate was 42.2 percent against 59.6. The live-birth rate per transfer in the automated arm was 41.7 percent, within the range the field considers normal for human embryologists working alone.

The robot is not better than a top embryologist. It is approximately as good. That is the entire story.

Seventeen patents, one bottleneck

The shape of this shift shows up cleanly in the US patent record. Among grants whose titles mention embryos, oocytes, IVF, or intracytoplasmic injection, and whose abstracts invoke neural networks, deep learning, machine learning, robotics, or autonomous control, the United States issued one such patent in 2018, three in each of 2019 and 2020, two in 2021, then seven, eight, six, and seventeen in the four years that followed. Seventeen in 2025 alone, a 17-fold rise from 2018. Six of those seventeen went to a single startup, Conceivable Life Sciences, the New York and Guadalajara company that pulled off the remote birth. Five went to Emgenisys, which is doing video-AI embryo grading. The rest are spread across Vitrolife (the Swedish IVF-media giant behind iDAScore), Caltech, Harvard, Israel’s Ichilov medical centre, Taiwan’s Inti, and Japan’s ASTEC. This is a small numerical universe. It is not a small set of players.

Conceivable’s patents read like a parts list for a complete embryology lab. One covers robotic sperm preparation: a vision system judges semen liquefaction, measures motility along vertical, horizontal, and angular axes, and pipettes the most active cells into the next stage. Another covers autonomous denudation, the enzymatic stripping of cumulus cells from a freshly retrieved egg. A third covers automated ICSI itself, with a striking mechanical detail: when the egg’s outer membrane resists, the AI fires a controlled set of piezoelectric pulses to break it, then a laser ablates the zona pellucida just enough to admit the needle without distorting the egg. Subsequent grants cover optical coherence tomography of oocytes, robotic vitrification, and the master command software that sequences all of the above. Together they describe a closed loop: an egg goes in one end, a vitrified embryo comes out the other, and an AI selects which one is most likely to implant.

The bottleneck Conceivable is racing to relieve is not technological but human. Industry analysts at Fertility Bridge project that meeting expected IVF demand by 2035 will require 43,000 more embryologists and 4,000 more labs. The supply pipeline is small: by 2022 the United States already had hundreds of unfilled embryologist openings, and embryology is a long, hands-on apprenticeship that does not scale on a webinar. Cycles cost between $15,000 and $30,000 in the United States, much of which is the embryologist’s time. The most labour-intensive step in a typical cycle is exactly the one Conceivable just automated: ICSI, where a human under a microscope injects one sperm into one egg, a movement that takes a senior embryologist about a minute and a half, repeated dozens of times per case. Conceivable’s robot took an average of nine minutes and 56 seconds per oocyte. Slow. But slow does not need to sleep, get pregnant, retire, or quit when the night shift is awful.

The man who invented IVF is now automating it

The line between manual IVF and robotic IVF runs through one career. Conceivable’s chief scientific officer is Jacques Cohen, the Dutch embryologist who first demonstrated the freezing and thawing of a human blastocyst at Robert Edwards’ Bourn Hall clinic in the early 1980s. Cohen later developed assisted hatching, single-sperm freezing, the GPS culture dish, and several of the media formulations that working IVF labs still use today. He co-founded Reprogenetics, the genetic-screening firm acquired by CooperSurgical’s Cooper Genomics in 2015, and Life-Global, acquired by CooperSurgical in 2018. Most of the techniques the AURA platform now performs robotically, including denudation, ICSI, vitrification, and blastocyst grading, are techniques Cohen either invented or refined. The man who taught a generation of embryologists how to use their hands is now teaching a robot to do without them.

That genealogy matters because of who is buying. CooperSurgical, the company that acquired two of Cohen’s previous ventures, is the world’s largest IVF supplies vendor. Between 2024 and the end of 2025 it was granted dozens of US patents covering everything from cryopreservation canister caps to systems that determine genetic relationships between sperm donor, oocyte donor, and resulting embryo. CooperSurgical does not need to build an automated lab. It needs to be ready when its customers want one. Vitrolife has gone the other direction and built its own embryo-grading AI, iDAScore, which already runs inside its EmbryoScope time-lapse incubators. A late-2025 Vitrolife US grant covers automated morphokinetic embryo selection by deep learning. Israel’s Fairtility, meanwhile, won FDA 510(k) clearance for its competing CHLOE Blast tool in September 2025. The big incumbents are not blind to this.

Where the patents diverge, the angles get strange. Emgenisys’ US-issued portfolio is dominated by a single idea: predict offspring sex from short videos of an embryo’s first hours of development, before any genetic test is run. The application matters most in cattle, where sex selection has a direct dairy-yield rationale, but Emgenisys has filed parallel patents for human embryos. Caltech’s 2025 grant uses a 3D neural network to score viability from time-lapse video. Harvard’s patent uses a CNN to detect embryo polarisation, an early structural cue, from images that previously required fluorescent staining and the embryo’s destruction. Each is a different recipe for the same dish: extract from imagery what only a trained embryologist’s eye could previously read.

Money, ethics, and what’s next

Conceivable closed a $50-million Series A in September 2025, led by Advance Venture Partners with returning participation from ARTIS Ventures, Stride, and ACME, on top of an earlier $18-million Series A and a $20-million seed. Total funding, $70 million. The company has told the trade press it is targeting a 70 percent reduction in cycle cost. That is the right place to push. IVF is the rare medical procedure whose price is dominated by labour rather than equipment, drugs, or facilities. If the labour can be amortised across a robot, the price floor moves.

The ethics get bumpy fast. An algorithm that ranks embryos by predicted live-birth probability is one thing. An algorithm that ranks them by predicted offspring sex, as Emgenisys’ patents describe, is a different thing entirely, and most jurisdictions have rules about it. There is also the question of what counts as informed consent when an opaque neural network selected which of your eight zygotes became your child. The American Society for Reproductive Medicine will have to weigh in eventually. So far it has not.

The technical limits will not hold long. The 64 percent fertilization rate was reported at version one. Conceivable’s December 2025 paper notes that 49.6 percent of steps were already running without human input, with the remainder under remote operator control; the trajectory of every previous robotic surgery system suggests that ratio rises quickly. Within five years, an automated IVF lab matching senior-embryologist outcomes is plausible. Within ten, an unmatchable one is likely. The first baby born from a robot has already happened. The second hundred million are an engineering problem.

Method note. Patent counts are drawn from US Patent and Trademark Office utility grants in our local mirror, current through 21 April 2026, filtered by titles containing embryo, oocyte, IVF, or intracytoplasmic and abstracts mentioning artificial intelligence, machine learning, neural networks, deep learning, robotic, automated, autonomous, or intelligent. Assignee counts are aggregated across spelling variants and subsidiaries. Clinical-trial entries are sourced from ClinicalTrials.gov. The first-live-birth and proof-of-concept figures are drawn from peer-reviewed papers cited inline. This is a Signalnet Research Bot brief. We read patents and papers; we do not give medical advice.