What it costs depends mostly on scope. Answering routine parts calls (phone rings, someone checks stock, quotes a price) is a narrower job than a system that also matches interchange (does this fender fit that other model, does this alternator cross over across model years), pulls a shipping quote for a bumper going out of state, and logs the exchange back into your inventory system. A single workflow costs less to build than a connected system covering inquiries, inventory, interchange, and shipping together. A full custom product, say a customer-facing quote tool with payments and multiple yard locations, is a different scale of build again.
The other big driver is whether you want it built and handed off, or built and operated. A one-time build is cheaper up front but leaves you to notice when it starts quoting stale prices, missing a new interchange match, or failing quietly on a busy Saturday. We run the systems we build: watching for failures, updating it as your inventory or pricing changes, and stepping in when something needs a person's judgment. That ongoing operation is part of what gets quoted, not something added later.
There is no fixed price list because the cost should track the value. Look at how much staff time goes into answering the same handful of question types, and how much of that can honestly be handed to a system versus needs a person's eyes on the part. If most incoming calls are "do you have this part" and "does this fit," that is a strong candidate. If most calls involve someone walking the yard to check condition or negotiating a price, less of the job can be automated, and the engagement should be scoped that way. The honest way to find out what it would cost for your yard is a short conversation about which of your calls are actually repetitive and which are not.