I remember well the meeting I had with a meat industry body some 15 years ago where they proclaimed no interest in IVF (then) because there was little to no genetic selection in beef breeds. The same industry body is now much more engaged. Yet the challenges of reduced embryo survival on transfer, along with the competence to perform vitrification well in the field, are still major problems for cattle IVF. This is where AI and automation could fit well, so long as the cost-benefit ratio is improved. The most significant opportunity is grading embryos prior to transfer, which is already clinically established, albeit with little evidence of improvement over a skilled clinical embryologist but is most likely faster than the embryologist. Pregnancy rates from cattle IVP embryos remain lower (50%- 55%) relative to MOET derived embryos (65%). In clinical IVF, the grading systems are aimed at selecting the best embryo, for the first fresh transfer, then the next (to vitrify), etc. This is because patients want to achieve a pregnancy as soon as possible. For cattle IVF, I have always maintained that we should be selecting for the embryos that have low to no chance of forming a pregnancy and exclude these and therefore train the AI systems to reduce the numbers of embryos. This clearly differs from the clinical scenario. However, there is an immediate data problem. Cattle IVF operations don’t normally take images of embryos, so where is the data going to come from to develop the algorithms? I have experienced the disruption in workflow that is created by having cattle embryo images taken before transfer. And the number of images to build and test algorithms is in the thousands. So pragmatism suggests that unless there is a major investment in time and effort by some (industry) group, we will continue to rely on the skill of embryologists to select cattle embryos for transfer.