Your dedicated style guru
Client
John Lewis & Partners
Year
2020
Tasks
User Experience
User Interface
User Testing
Prototyping
Details
The style bot is a tool that allows customers to set their budget, style preferences and measurements in order to get a curated outfit. The output is then shared with an AI driven styling tool, to help human stylists pick 5 pieces of clothes tailored for the customer.
The bot would be integrated with WhatsApp and Facebook chat so that customers wouldn’t need to download a special app.
Understanding style
Working closely with the Tech Lead, we captured the different layers of our data model to get us closer to understanding the customers’ personal preferences.
One approach was finding people with similar wardrobes. We would then assume that customers with similar taste would share similar recommendations.
Another approach was classifying people into style tribes. A style tribe would have different facets, called families and there would also be a final layer of personal preference.
However, this approach would require detailed understanding of garment characteristics, which we didn’t have available at that point.
Dialogue tree
WhatsApp integration - user flow
Recognising intention
We built the bot through Dialogflow. It uses natural language processing and machine learning.
For each task, we had to specify intents (customer objectives) and entities (variables) so that the system could process the customer’s answers and reply accordingly. Fallbacks were also created in case the bot couldn’t understand the customer’s intents.
