The call that actually eats the counter is a lookup: do you have a driver door for a 2014 Silverado, what does it cost. An agent can sit on that line (and on texts), pull the answer from the inventory system a scrap yard or auto salvage yard already uses, and draft a ticket so the ask is logged instead of living in someone's head. It answers in seconds, after hours or during lunch, and it does not forget to ask for the VIN. Interchange numbers and a shipping quote by weight and destination are the same kind of lookup, added once the phone-and-ticket path is stable. The answer already exists in a system. The agent fetches it and writes the ticket.
What it cannot do is look at a part and tell you if that door has hail damage, whether a wiring harness was cut during the pull, or whether a customer's oddball fit will actually work. It cannot negotiate price with a rebuilder, and it should not be the one denying a core return. Those are judgment calls. A system that pretends otherwise costs you trust with a regular buyer. We map the manual work before building anything, so we can say which of your counter work is a lookup that can be automated and which is a judgment call that cannot.
In practice this runs as an operations system: it logs every inquiry as a ticket, hands off anything it cannot answer confidently to a person instead of guessing, and treats inventory data as the single source of truth so it never quotes a part that already sold. If you also want customers to search and pay for parts without calling in, that is a bigger step, a small production web app with real search and payment. Either way we build it, deploy it, and keep running it, watching for wrong answers or missed handoffs, instead of handing you a system and walking away. What a phone-and-ticket workflow actually costs to build and run depends on how much of the inquiry volume you want covered and how many systems it has to connect to. We quote per engagement.