Automating customer support via AI-powered chatbot
A go-to brand for home furnishing in the USA
The brand is one of the leading distributors of furniture, carpets, curtains, and other home decor accessories. With its presence in the USA and across the globe, the brand owns multiple sub-brands in the home furnishing industry, all known for their high quality, fashionable design and luxury.
Amassing a massive list of products under several home decor categories, the brand had to cater to thousands of user inquiries.
The brand was looking to automate the responses, minimizing the support team’s responsibilities and allowing them to focus on more important work.
The brand was looking for Artificial Intelligence experts to build a mechanism that can handle most of the frequently asked user inquiries.
Ziffity’s AI experts organized brainstorming sessions with client-side stakeholders to understand the areas where the brand is trying to automate responses and the workflow of the existing customer support interactions.
Analyzing the challenges the brand was trying to solve, team Ziffity zeroed in on building a Chatbot using Google’s Dialogflow. This platform helps design conversational UI on mobile, web, messengers, and smart devices.
Chat Script for the Workflow
Based on the understanding from the discovery phase, team Ziffity started sketching a conversational workflow for the use cases ‘Order Status’ and ‘Stock Availability.’ While determining the chat flow, Ziffity also addressed the exceptions to be taken care of.
Based on the script developed, team Ziffity created multiple intents for each use case to ensure that the chatbot understands different phrases used by customers while interacting. Team Zififty also defined the entities and context to make the interactions with the bot more human-like.
Ziffity integrated the chatbot with the client’s ERP (through API calls) system to fetch stock and order data. To confirm that the data is relevant to the user interacting with the chatbot, team Ziffity facilitated confirmation layers that match users' Order ID and Email ID.
Zififty’s QA experts performed testing across the data exchange between the chatbot, eCommerce backend, and the ERP system. User acceptance testing ensured that the chatbot UI enabled users to complete their actions.