Ask an AI coding assistant for “a 3D product configurator for my chair” and within minutes you’ll have a page with a rotating model and a few colour buttons. It feels like the problem is solved. For a demo, it is. For a shop, that page is maybe 10% of the work.
Short answer: use AI to speed up the parts around your configurator, and use a maintained plugin for the part that touches your orders. Here’s why.
What AI is genuinely good at
- Prototyping the look. A three.js or model-viewer page that swaps colours is a classic task, and AI does it well. Great for testing an idea with your team.
- Preparing 3D assets. AI can write Blender scripts that rename hundreds of objects, batch-export variants or generate texture variations. Text-to-3D and image-to-3D tools can produce rough models for simple products.
- Writing content. Option descriptions, care notes, material explanations, alt text, translations.
- Planning your rules. Describe your product and ask AI to list which options conflict. You get a first draft of your conditional logic to check with your workshop.
- Small custom code. A WooCommerce hook or filter to tweak one behaviour, on top of a plugin that already does the heavy lifting.
Where an AI-built configurator gets hard
The visible part of a configurator is the easy part. The rest is what keeps orders correct.
- Cart and order data. Every choice must reach the cart, the order, the confirmation email, the invoice and the admin screen, in a format your team can build from. WooCommerce has many edge cases here (sessions, HPOS order storage, REST, block checkout).
- Validation. A customer must not be able to order a battery that doesn’t fit the frame. Rules have to be enforced on the server too, not just hidden in the interface.
- Performance on phones. Lazy loading, model compression, texture sizes, memory limits on iOS Safari. AI-generated demos usually load everything at once.
- Editing without a developer. Next season you add a fabric. With a custom build, that’s a code change. With a plugin, it’s a few clicks in an editor.
- Maintenance. WordPress, WooCommerce, three.js and browsers all ship updates. Somebody has to keep the code working, and “the AI wrote it” doesn’t help when checkout breaks on a Saturday.
- Security. File uploads, custom fields and AJAX endpoints are classic attack surfaces. Generated code often skips nonces, capability checks and sanitisation.
The real cost comparison
| AI-built custom configurator | Configurator plugin | |
|---|---|---|
| Time to a working demo | Hours | Hours |
| Time to a production-ready shop | Weeks of developer time | Days, mostly preparing the model |
| Who edits options | A developer | You, in the editor |
| Price and rule enforcement | You must build and test it | Built in |
| Updates and compatibility | Your responsibility | The plugin’s |
| Cost | Developer time + ongoing maintenance | Free core, optional add-ons per year |
Custom still makes sense in some cases: a product unlike anything else, a configurator that must talk to a custom ERP or a CAD system, or a team with in-house developers who will maintain it for years. Even then, many agencies start from a plugin and extend it with hooks rather than starting from zero.
A practical way to combine both
- Use AI to help prepare and optimise your 3D model (Blender scripts, naming, compression settings).
- Build the configurator in MKL Product Configurator for WooCommerce, free, in 2D or 3D.
- Use AI to write option descriptions and draft your rules, then enter them in the editor.
- If you need one behaviour the plugin doesn’t have, ask AI for a small hook or filter, and have it reviewed before it goes live.
- Creating a custom Configurator theme for the MKL Product Configurator for WooCommerce – provide the Custom theme documentation to the AI together with your design to create a working configurator theme
You get the speed of AI where mistakes are cheap, and proven code where mistakes cost you orders.
The verdict
Should you use AI to build a custom product configurator? Use it to build a prototype, prepare assets and write content. Don’t rely on it alone for the part that takes money and creates orders, unless you have a developer who will own that code long term.
Try the free configurator · See the demos
FAQ
Can ChatGPT or Claude build a product configurator?
They can build a working visual demo quickly. Connecting it safely to cart, pricing, orders and stock takes much more work and testing.
Can AI generate the 3D model or layers for my product?
For simple shapes, AI 3D generators give a usable start. For products you sell, you usually still need accurate dimensions and clean, separate parts, so expect manual clean-up in Blender.
Is a plugin less flexible than custom code?
A good plugin exposes hooks and filters, so developers can extend it without forking. This is where AI shines most
