Computer vision and other AI technologies are quickly growing as a tool for entomologists. Classification and object detection algorithms, for example, are enabling the rapid counting and identification of different insect species, including bees. With continued refinement, we will soon begin relying on these algorithms for a much greater proportion of our research and conservation programs. It has become apparent that the algorithms perform very well but are mainly limited by the availability of high-quality training data. Although public repositories like GBIF provide millions of useful images from a variety of contributors that are useful, the vast majority of bee species are not well represented. We are therefore exploring how to build high-quality image datasets that can be applied to a variety of classification tasks and placed in the field on a variety of devices. Users will be using the technology different purposes that require different types of hardware. This could be mobile or web app based, placed in camera traps, or placed on mobile and autonomous vehicles. The algorithms will need to be nimble and flexible to work with sensors in these different contexts. I will discuss some of my lab group’s research in this area, especially our work on incorporating the efficient gathering of different image types and how our algorithms perform in different entomological contexts.