Intake starts wherever customers already reach you, a web form, an email inbox, a chat widget, even a shared spreadsheet. The agent reads each incoming order, pulls out the specifics it needs (size, material, quantity, deadline, whatever your product requires), and turns it into a structured record. If something is missing or contradictory, it writes back to the customer asking for that one thing, instead of handing your team a half-finished order to chase down.
It then checks the order against rules you set: material stock, capacity for the promised date, price range, minimum order size, whatever actually decides yes or no. Orders that clear every check get confirmed and pushed into your production queue on their own, with a confirmation sent to the customer. Orders that don't, a spec you've never quoted, a deadline you can't hit, a price under negotiation, get routed to a person with the order and the reason attached, so the only thing left to decide is the part that actually needs a person.
This fits a business where most orders are variations on things you already know how to make and price. It fits less well where almost every order is a real design consultation, since then the system is just moving paperwork around a decision someone still has to make from scratch. We connect it to whatever you already run orders through, a shared inbox, Airtable, Shopify, a workshop tool, rather than asking you to switch systems, and if you eventually need customers placing and tracking orders themselves, that becomes a proper web app rather than an add-on. Start by counting how many of last month's orders needed a real judgment call versus how many were routine: that ratio tells you how much of this can run on its own.