Ai Assistant

B2C Modular Configurator plugin that allows a user to configure their perfect product by using Ai prompts or image recognition

Overview

Looking to join the race to introduce Ai into our product set and continue to be market leading we looked across our products and identified areas where Ai could help. Taking into account our current limited knowledge in the Ai space and the complex nature of our products we opted for a client facing Ai integration that looked to tackle the overwhelming configuration options available in our customers modular products.

Goal

Develop a straightforward and user-friendly AI interface that allows a user to easily supply a simple text prompt, and in response, the AI will intelligently construct a product that effectively suits their specific needs and requirements. This interface should be designed with accessibility in mind, ensuring a seamless experience for all users.

Deliverables

Output

Ai integrated component

New Icon set

Ai data capture and learning

Chat to ai response flows

Component to output UX

Effect

Add to cart rate increased by 43%

Ai adoption 86%

PDP dropout decreased by 17%

Just for context

Here is a live example of the modular configurator application deployed for Design with in reach (taken 10.01.2025). A US home and office furnishing customer providing direct customer access via their e-commerce platform.

Wireframes

Steps I took

  1. Persona creation

  2. Analytic data collection and analysis

  3. Workshop with client

  4. Customer survey

  5. Competitor analysis (new space)

  6. Site map updates

  7. PM/PO and Dev alignment

  8. Ai data learning process

  9. Existing component check

  10. Wireframe creation

  11. Wining designs refinement

  12. Prototype creation

  13. Live tool monitoring

  14. Live product survey

  15. Product use video

What I learned

Before live date

  1. Ai needs to be given data to learn from and will only get better over time. This is a lengthy process.

  2. Ai needs to be fed product data and persona data to know how to respond.

After live date

  1. Users create their product 60% faster with ai then when navigating the menus.

  2. Users engaged more with the ai after seeing a video of how it works.

  3. Users look through 12% more configuration options before buying when using the ai over the existing menu options.

  4. 73% of users that took the live product survey said they were more likely to recommend the brand after interacting with the ai tool.

Takeaways

  • Some users stated that their fear of ai delayed them using the tool but once they tried it wished that all shopping experiences matched this one.

  • User adoption went down slightly when I added the term ai into the tool but return visits went up.

  • Users commented on how fast the tool was and enjoyed the unscripted interaction.

  • This ai tool needs no alteration for use on the mobile experience.

  • Allowing the user to speak to the ai with voice notes and add contextual images provides a new layer of power to the tool that sets it apart from other live experiences (Feb 2025)

Final Design

Simple user flow

Summery

This early step into ai for 3D Cloud showed to their customers that they are at the forefront of technology development and their quick time to market at only 6 weeks design/development time showed that we are looking to get tools in front of live users to gain real insight.

In this project we discovered that training an ai to provide meaningful results takes time and patience. Our planning and research saved us a lot of time and allow us to feed the ai a huge amount of meaningful data. However my team still chose to step up and assist the data BA’s and QA’s in training the ai to ensure users receive meaningful results lending to the projects success.

Given the chance to do this project again I would get a build up sooner, with an informational notice letting user know that we are testing the ai and have real user data aid in its training.