Many businesses do not sell a fixed list of items. A customer picks a color and size from products already on the shelf, or chooses the parts of a product that have not been built yet. Every one of those choices has to land on the right product, the right price, and the right stock record, and for made-to-order products, on the right steps on the shop floor.
Apache OFBiz, the open-source ERP from the Apache Software Foundation, handles both cases. Ready-made choices, like sunglasses in different colors and widths, are set up as product variants. Choices that decide what gets built, like the frame, lens, and coating of custom glasses, are set up as configurable products. Doing this by hand means connecting products, prices, stock, and assembly steps correctly, and a single missing link can stop the setup from working.
This tutorial shows a faster way. You describe your business in plain language, an AI assistant turns it into Apache OFBiz data, you review what it understood, and you import one file. Below, we cover what the setup involves, how to build it with AI, and what to check before deciding whether Apache OFBiz fits your business.
It takes more than creating a product and adding a few dropdowns. Product variants, configurable options, pricing, store and catalog relationships, inventory, and manufacturing data need to agree with each other.
Product variants. These are predefined versions of a product, such as sunglasses in different frame colors and widths. Each valid combination must resolve to a specific product with its own price and inventory records. Apache OFBiz uses product features and product relationships to represent these combinations. If a combination is missing or incorrectly linked, the storefront may not offer the variant you expect.
Configure-to-order products. Here, customers choose options that define the product they want made. For example, a customer might select a frame, lens type, and protective coating for custom glasses. Each option must link to the actual component behind it, and the questions should follow the order in which the product is assembled. A dropdown alone is not enough if the selected option does not resolve to a component product.
Pricing and stores. The displayed price must reflect the selected variant or components. The product store also needs to be associated with the right catalog, and the website needs to point to the intended store. If these relationships are wrong, prices may not resolve, or the storefront may appear empty even when the data was imported.
Inventory and assembly. Stock should be assigned to the variants and component products that are actually held in inventory. A made-to-order configurable product may have no stock of its own because it is assembled after an order; its components should be stocked instead. If the product is assembled to order, its routing must also be linked to the product so the order can create the expected production run.
None of this is unusual. The challenge is getting the relationships right. A successful import is only the first check; the storefront and order flow still need to be tested.
This is the short version of the setup. We have documented the complete, detailed process on the Apache OFBiz community wiki, including the business questions, the full prompt, import instructions, verification checks with screenshots, and troubleshooting. The link to the full guide is at the end of this section.
1. Describe your business. Include your product families, variants, configurable options and their components, prices, markets, stores, warehouses, starting stock, and assembly steps if you make products to order. You can answer the questions in the guide or describe your business in a paragraph.
"We sell sunglasses and prescription glasses. Customers can choose frame color and width for our ready-made sunglasses. For custom glasses, they choose a frame, a lens type, and an optional protective coating. We stock frames and lenses at our central warehouse and assemble the glasses after an order is placed."
Missing details are filled with defaults and marked as assumptions for you to review.
2. Download the prompt file: master-prompt-configurable-product.txt.
3. Give the prompt and your business description to an AI assistant. Follow the prompt's instructions. If the assistant cannot run code, the guide explains how to get the structured specification and run the supplied generator on your own machine.
4. Review the specification. The assistant marks each detail as either “given” or “assumed.” Review the assumptions carefully. You might need to add a frame size, change a base price, or remove a store you do not need. When the specification is right, type GO to proceed.
5. Generate and import. Import the generated XML through WebTools, leaving Create Dummy FKs and Disable ECAs unchecked. Apache OFBiz should report Got N entities to write to the datasource. Compare that count with the generator’s report, and clear all caches if the new data does not appear.
For the complete, detailed setup, follow the full guide: Set Up Your Configurable Product Business in Apache OFBiz with AI
A working starting point for testing variants and configurable products, not a finished production storefront. The generated setup can include:
After importing, check that each choice resolves to the right product and price, and that an approved order creates a production run with the parts the customer chose. The full guide on the wiki covers every check with screenshots.
The customer's choices carry through to manufacturing. The approved order created a production run with the three assembly steps and the exact parts selected.
The generated data gives you a working basis for testing, but some parts still need business-specific configuration.
If your business handles products with a shelf life, such as food or beverages, there is a companion guide for setting up perishable inventory:
Set Up Your Perishable Goods Business in Apache OFBiz with AI
The point of this setup is not just to put variants and options on a product page. It is to find out whether Apache OFBiz can represent the choices your customers make and connect those choices to the right products, prices, inventory, and assembly steps.
Use the generated configuration to test your own catalog. Do variant combinations resolve to the right products? Does the price change when a customer selects a different component? Does the order carry those choices through to the production run?
If the basic model fits but your pricing rules, product logic, or order-to-manufacturing flow needs more control, those gaps may call for additional implementation. That is the kind of work HotWax Systems helps businesses evaluate and build. Tell us what your configurable-product workflow needs, and we can discuss what it would take.