Buying a bra online carries a little bit of an act of faith, this step process gets more difficult for the sports bra category: the sizing, fit, and texture are all highly dependent on a number of variables, somewhat unique and subjective to each consumer. Jockey wanted to tap into their vast knowledge base of how their underwear works toward guiding customers to the best garment. But this involves multiple data dimensions and could be a fairly tricky navigation through tables and multiple questions.

We built this AI-powered interactive app on our low-code development platform Interplay. The flow diagram is shown above.  To build this app, we dragged-and-dropped those nodes together, configured each one, and then hit the ‘deploy’ button.  Yeah, that quick. Among the nodes used were the following:

• Algolia Search/Index: Searching and Indexing questions
• Interplay NLP: module to train and detect Intent
• HTTPIn/HTTOut Adapters create the REST calls

To address this challenge, Iterate designed an AI-based natural language FAQ: customers can ask a question in normal conversational English as if they were speaking to a sales clerk in the store. The AI returns a recommended product based on the best fit to address the customer’s question. Note: many common recommendation engines will send the ‘most popular’ or ‘best selling’ product, which is exactly what Jockey wanted to avoid.

To set up this AI-powered FAQ, our team did the following:

• Spin up an instance of Interplay to prototype an AI engine with the training data (the knowledgebase from Jockey that has body size information, product fits, etc.).
• Train the AI to understand common-language questions by utilizing known NLP methods that already exist in Interplay
• Judge the results, repeat through several epochs to improve the quality of answers
• Connect the smart FAQ to a JSON output for simple plugging into the jockey website
• Launch into production!

With this FAQ in place, Jockey saw an increase north of 50% in conversion for shoppers who utilized the FAQ!

To find out more how we can bring similar smarts to your website and product information on behalf of your customers, contact us below: