# Alan Murray, Co-founder and CEO of Conceivable Life Sciences – Interview Series > Source: > Published: 2026-07-31 11:25:35+00:00 ### [ Interviews ](https://www.unite.ai/series/interviews/) # Alan Murray, Co-founder and CEO of Conceivable Life Sciences – Interview Series [Add Unite.AI to your preferred sources on Google](https://www.google.com/preferences/source?q=unite.ai) [Alan Murray](https://www.linkedin.com/in/alanleroymurray/), Co-founder and CEO, Conceivable Life Sciences – is an engineer and serial entrepreneur with more than three decades of experience spanning technology, machine learning, hardware, logistics and company building. Before launching Conceivable, he co-founded TMRW Life Sciences, where he helped develop an automated platform for managing and safeguarding frozen eggs and embryos used in in vitro fertilization (IVF). Murray is also a co-founder and partner at Coriolis Ventures and has helped establish companies across digital advertising, hospitality and life sciences, including Integral Ad Science ([IAS](https://www.securities.io/nasdaq/IAS/) ), Dstillery and NeueHouse. At Conceivable, he serves as both a strategic leader and the chief architect of AURA, overseeing the integration of robotics, advanced imaging and machine learning into fertility laboratory workflows. [Conceivable Life Sciences](https://www.conceivable.life/) is a biotechnology company developing AURA, an automation-assisted IVF laboratory designed to coordinate the complex processes involved in creating, cultivating and preserving embryos. Rather than automating a single procedure, the platform integrates robotics, software, advanced optics, artificial intelligence and laboratory equipment across the IVF workflow, while maintaining digital traceability and allowing embryologists to retain control over clinical decisions. Conceivable aims to use this standardized approach to reduce variability, expand laboratory capacity and make fertility treatment more accessible and affordable to families who currently face high costs or limited access to specialist care. **You previously co-founded TMRW Life Sciences, where you helped introduce automation into the storage and management of frozen eggs and embryos. What did that experience teach you about the limitations of today’s fertility infrastructure, and what ultimately convinced you to found Conceivable Life Sciences and tackle the more complex challenge of automating embryo creation?** At TMRW, we automated how eggs and embryos are stored and tracked once they exist. It solved a real problem, chain of custody in cryopreservation is unforgiving, and getting that right matters enormously to patients. But it also showed me the limits of what storage automation alone could fix. The bottleneck in IVF was never really what happens after the embryo is made. It’s the making of it. That work still depends on a small number of highly trained embryologists doing manual, repetitive, sub-micron precision tasks by hand, and the outcome varies by who’s holding the pipette that day. When I started TMRW, the fertility storage market was still treated as a niche corner of reproductive medicine. I didn’t see it that way. Delayed parenthood was becoming the norm, more people were freezing eggs and sperm earlier, employers were starting to add fertility preservation to benefits packages, and the LGBTQ+ and single-parent communities were building families through egg and sperm banking at a scale the industry hadn’t planned for. All of that pointed to a storage and tracking problem that was about to get much bigger, much faster than most people in the space anticipated. That’s part of why chain-of-custody mattered so much to me at TMRW. If you’re building for a market headed toward that kind of scale, you can’t rely on manual tracking and human memory to keep eggs, sperm, and embryos straight. Automating storage and tracking wasn’t just about doing today’s job better, it was about building infrastructure that could hold up as the number of patients, cycles, and specimens grew by an order of magnitude. In 2018 I started to design what became AURA from the ground up, from microfluidics to robotics integration to injection technique. What convinced me to create Conceivable was pretty simple: if we could bring the same rigor to embryo creation that manufacturing and robotics have brought to other precision industries, we could make IVF more consistent and more accessible at the same time. That was the harder problem, but it was the one worth solving. **AURA coordinates more than 200 steps across interconnected systems for dish preparation, sperm processing, egg identification, intracytoplasmic sperm injection, incubation, robotic transport, and vitrification. How do the software, robotics, imaging systems, and laboratory equipment operate as a unified platform while maintaining traceability for every egg, sperm cell, and embryo?** AURA runs these more than 200 steps as one continuous system rather than a chain of separate machines. The software layer is what makes that possible. Every action, every image, every measurement is tied to a specific egg, sperm cell, or embryo through the entire cycle, so there’s a full digital record of exactly what happened and when. That traceability is the part people underestimate. In a conventional lab, a lot of that history lives in an embryologist’s notes and memory. In AURA, it’s structured data from the first step to the last, which is what lets the robotics, imaging, and lab equipment coordinate in real time and lets an embryologist step in at any point with full context, not a partial picture. **Your technical stack includes a Detection Transformer to assist with sperm identification, advanced imaging for locating eggs, and vision-language-action models designed to guide delicate physical procedures. How are these models trained, what signals do they evaluate, and how does the system determine when an embryologist needs to intervene?** We use a Detection Transformer to help identify and track sperm, computer vision models to locate and assess eggs, and vision-language-action models to guide the physical precision required during procedures like ICSI. These models are trained on large sets of labeled cycle data, image sequences paired with outcomes and embryologist annotations, so the system learns to recognize the visual and procedural signals that matter clinically, not just what a “normal” cycle looks like on average. Every clinical decision point in AURA is designed with a human checkpoint. The system is built to flag uncertainty rather than push through it. If confidence on identification or a procedural step falls outside the range we’ve validated, it stops and calls in an embryologist. AURA is automation-assisted–and that’s a deliberate design choice, not a limitation we’re working around. **You have drawn inspiration from semiconductor manufacturing and autonomous vehicles, industries that combine perception, planning, and precise physical execution. Which engineering principles transferred successfully into embryology, and which had to be redesigned because AURA is manipulating living cells rather than manufactured components?** The engineering principles that transferred well were around precision manufacturing at scale: tight process control, sensor fusion, treating variability as something to measure and design out rather than accept. Autonomous vehicles gave us a good model for how perception, planning, and execution need to work together in real time, and how a system should behave when it isn’t confident about what it’s seeing. What had to be redesigned almost entirely was our tolerance for the unexpected. A semiconductor wafer behaves the same way every time. A living cell doesn’t. Two eggs from the same patient can look different, behave differently, and respond differently to the same procedure. So instead of building toward a single “correct” output, we had to build a system that can recognize a wide range of biological variation and knows the difference between normal variation and something that needs a person to look at it. That’s a fundamentally different engineering problem than anything in chip manufacturing or driving. **Conceivable describes each AURA cycle as contributing to a data flywheel of standardized process information that conventional laboratories cannot easily collect. What data is captured during a cycle, how can it improve future models, and how are you addressing patient privacy, demographic bias, differences between clinics, and the risks of updating clinical AI systems?** Every AURA cycle generates a standardized dataset: images, sensor data, timing, and outcomes tied to each step, in a level of detail that manual IVF labs simply don’t produce today because so much of the process isn’t recorded systematically. That data feeds back into improving our models over time, which is one of the real advantages of automation. The system gets better with volume in a way manual processes can’t. On privacy, patient data is de-identified for model training and handled under the same regulatory standards that govern clinical data generally. On bias, we’re deliberate about training on data across different patient populations and clinical sites so the models don’t just perform well for the demographics best represented in early data. Differences between clinics are a real variable we track for, since lab conditions and patient populations vary. And on updating clinical AI systems safely, any model update goes through validation before deployment, and we’re not moving faster than the evidence supports on that front. We’d rather be right than first. **Conceivable has conducted pilot studies involving more than 100 patients and over 1,000 eggs. Which clinical endpoints will be most important in proving that automation can improve consistency, safety, embryo development, pregnancy rates, or live-birth outcomes compared with conventional manual IVF?** With more than 100 patients and over 1,000 eggs in our pilot work, the endpoints that matter most are the ones that actually predict a healthy outcome for a patient, not just process metrics. Fertilization rate, embryo development to blastocyst, and euploidy rates tell us whether the biology is behaving the way it should. From there, implantation and pregnancy rates tell us whether that translates into a real shot at a baby. Live birth is the endpoint that matters most, and it’s also the slowest to accumulate, since it takes months after transfer to know. What we’re most focused on proving is consistency: that AURA produces the same quality outcome regardless of which lab, which day, or which patient population it’s serving. That’s the piece manual IVF has never been able to guarantee, and it’s the piece automation is built to solve. **C:VIT is designed to cool embryos up to 50 times faster than conventional vitrification methods while standardizing cryoprotectant exposure and reducing ice-crystal formation. How does faster cooling translate into clinically meaningful benefits, and what evidence will be needed to demonstrate improvements in embryo survival, implantation, and live-birth rates?** The clinical logic is straightforward: slower cooling gives ice crystals more time to form, and ice crystals damage cell structure. Faster, more standardized cooling reduces that risk and gives every embryo the same cryoprotectant exposure, rather than exposure that varies by which embryologist is performing the procedure and how quickly they work. We designed C:VIT to improve embryo survival through the freeze-thaw cycle, which is the first and most immediate thing to measure. From there, the evidence that matters is implantation and live birth rates in embryos vitrified with C:VIT compared to conventional methods. We’re building that evidence deliberately, through controlled studies, before making claims about outcomes at scale. **Conceivable emphasizes that AURA is automation-assisted and that clinical decisions remain with embryologists. Which decisions must always remain under human control, what safeguards allow an embryologist to stop or override the system, and how is accountability handled when an AI recommendation or robotic action affects an egg or embryo?** Every decision that affects clinical judgment, whether an egg is viable, whether an embryo is suitable for transfer, whether to proceed with a procedure, stays with the embryologist. AURA is automation-assisted. It’s designed to execute physical tasks with precision and consistency and to flag anything outside expected parameters, but it doesn’t make clinical calls on its own. Embryologists can stop or override the system at any point in a cycle, and the system is designed to make that easy, not something you have to fight the interface to do. On accountability, every action AURA takes is logged and tied to the embryologist overseeing that cycle, so there’s a clear record of what the system did, what it flagged, and what decision a human made in response. That traceability is what makes accountability possible in the first place. **Conceivable and IVI RMA plan to deploy the first AURA system at a United States clinic in 2027. What regulatory, clinical-validation, infrastructure, and workflow challenges must be resolved before the platform can move from controlled studies into routine patient care?** There’s real work ahead on four fronts. Regulatory: working through the appropriate pathway with the FDA for a system that combines robotics, software, and reproductive medicine, which doesn’t have a lot of precedent to follow. Clinical validation: building the evidence base at the scale needed for a US clinical setting, beyond what pilot studies can show. Infrastructure: standing up the physical deployment at a US clinic with IVIRMA, which is its own significant undertaking. And workflow: making sure embryologists and clinical staff at that first site are trained and comfortable with the system before it touches patient care. None of those are small, and we’re not treating 2027 as a date we hit regardless of where the evidence lands. It’s the date we’re working toward when the validation supports it. **IVF remains expensive, geographically limited, and dependent on a relatively small number of highly trained specialists. What must change in the economics and operating model of fertility care for automation to make IVF more accessible, and how will you measure whether AURA is genuinely reducing the cost per successful birth?** The core problem is that IVF has depended on a small number of highly trained specialists doing manual work that doesn’t scale. That’s what makes it expensive and geographically limited. It’s not a lack of demand, it’s a bottleneck in supply. We recently co-authored a study looking at cost-to-baby as a share of household income across 25 countries, and the pattern is stark. Countries where the net cost a patient pays is below roughly half of median household income, Israel, Japan, Spain, Taiwan, see IVF account for something like 9 to 12 percent of all births. In the U.S., where net costs typically run above 75 percent of household income, that number is about 2.5 percent. Affordability isn’t one factor among many. It looks like the primary barrier. If the U.S. saw similar gains to what the study found elsewhere, that could mean something on the order of 140,000 to 160,000 additional births a year. Behind that number are families who want a child and can’t currently afford the path to one. Automation addresses that bottleneck directly: a standardized platform can bring lab-quality precision to more locations without requiring the same concentration of rare, highly trained talent in every clinic, and fewer cycles per successful outcome is the lever that actually moves cost. For measuring whether AURA is genuinely reducing cost per successful birth, the number that matters isn’t cost per cycle. It’s cost per live birth, since a cheaper cycle that doesn’t work isn’t actually more affordable for a family. We’re tracking that metric directly as we scale, and we’d rather be conservative about what we claim until the data supports it. *Thank you for the great interview, readers who wish to learn more should visit Conceivable Life Sciences.*