Being the first Pharmacy in the UAE to go the eCommerce way, Aster Online is the eCommerce arm of the Aster Pharmacy business; wholly owned by Aster DM Healthcare. Backed with more than 3 decades of experience in pharmaceutical retailing, Aster Online is designed to be a one-stop-shop offering premium healthcare and wellness products.
With over 7,000 SKUs across various beauty and wellness categories, the Aster Online webshop offers its customers the ability to browse through products via search filters, or even chat with a live pharmacist.
Client:Aster Primary Care
Key Focus:NLP chatbot
Geared with a backend contact centre team of 20 Pharmacists who handled all product queries coming through phone calls, their website (via live chat) and WhatsApp (without a verified Business Account); their response wait-time stood at about 5 minutes. This was extremely frustrating for their customers.
Their key challenges included:
Reducing the burden on the contact centre
Bi-lingual support, 24/7
Omni-channel presence: Website and WhatsApp (via the Business API)
Another challenge was the unavailability of APIs to the backend product database or web services, hence seeming to be a major show-stopper.
A Discovery workshop with the Business Head and other key stakeholders helped us identify the below use-cases:
Our team designed and developed a hybrid chatbot, driven by Conversation AI and NLP (Natural Language Processing). The chatbot covered the key business use-cases identified via the Discovery Workshops detailed above.
Our Solution Architect devised a workaround to solve the API challenge, or the lack of it, specifically. This included the development of a SQL database hosted on the same AWS server as the product & the frontend. For the client to manage and update data in real-time, a simple google sheet was created to update all product information and also add new ones. We then ran a python script that parses through the entire spreadsheet of 7000 products and thereafter updates the SQL database.
Entity extraction also played a crucial role in the development of this bot, due to the extensive range of products & the various sub-categories that fall under them. The advantage our development team had were our eCommerce pre-trained data sets & NLU library which helped in making this bot robust.
The solution included the below key customisations to their bot: