Founded in 2019, San Francisco-based Sewer AI has raised $2 million in disclosed funding from a Seed round in August 2020. The company uses a computer vision platform called AutoCode to assess defects in sewer infrastructure from video footage autonomously. Currently, robots are used to crawl sewers and image the pipes. Humans must then watch the resulting video – every darn boring foot and excruciating second in real time – and jot down any problems they see. The system is particularly adept at detecting what are called cross bores, or unintended intersections of utility lines, which could compromise the integrity of either or both.