When he left Robinhood, the author spent months working inside small businesses—first a Menlo Park dental office, then a gastroenterologist’s practice in Scranton—to see how much time went into paperwork. He found his dentist logging 2,400 hours a year on admin. With over half a million medical offices and countless other small businesses spending roughly $200,000 annually on staff for billing, scheduling, claims and data entry, these owners end up doing the work themselves. They’d rather treat patients or fix sinks than click through insurance portals, reconcile payments by hand or batch-print statements. AI agents change that. They can move across systems, interpret messy data and handle repetitive tasks end to end. The challenge is building a system reliable enough for regulated fields—one that files claims correctly, updates practice records, talks to insurance portals, clears payments and even talks to patients. It must juggle dozens of APIs and failure modes and install in minutes so busy, nontechnical users adopt it without hand-holding. The team built “Lassie,” an autonomous admin assistant now live in over 700 practices across 49 states. On average it saves each office 30 hours of work per month; some get up to 190 hours back. The founders became experts by doing the work themselves—reconciling millions in claims and billing thousands of patients—then onboarded early customers in person, watching over their shoulders from Florida to Oregon. That hands-on approach helped them refine the install flow, link bank accounts and record systems, and prove the AI’s reliability. Doctors using Lassie now see more patients, leave the office on time and even take vacations. One said it saved his marriage. The product’s rapid word-of-mouth growth shows how much small-business owners will embrace AI when it truly works and frees them from busywork.