Nicholas Wodtke
CEO
Vitaminology
Vitaminology uses AI and image recognition to fetch and present data to create the first search engine for vitamins and supplements.
-
Country
United Kingdom
-
Industry
E-commerce / Healthcare
-
Duration
Ongoing
Complex Data Infrastructures
ML Application Development
Overview
Vitaminology is the world’s first dedicated search engine for vitamins and supplements. The Vitaminology platform enables its customers to search for, discover, and compare supplements, empowering them to make better-informed self-care and wellbeing decisions.
A key differentiator for Vitaminology is that it serves as a single source of information for its customers, saving them the time and hassle of visiting multiple different retailers, finding the right products (they could be listed in different categories), and displaying the most-accurate details for a given product, such as price.
Overview
42,000+ products
in the database
2-week
release schedules
60+ microservices
integrated
Challenge
The Vitaminology platform relies on its ability to filter through and display large amounts of constantly changing data to ensure members can access up-to-date, relevant information about health products.
The challenge faced by Vitaminology was how to build and continuously maintain a database of thousands of health care products displaying different information in different formats while managing the complexity of structured and unstructured data (text-based information, marketing material, regulatory information, product label images, etc.).
Challenge
Solution
The Cledar team began by analyzing the unique needs of Vitaminology. A decision was made to create a solution from scratch. This custom software solution would be designed to reflect the unique characteristics of different websites, product categories and brands.
We then developed a dedicated system using advanced image recognition and Optical Character Recognition (OCR) to analyze the product labels shown in photos and images, as well as Natural Language Processing (NLP) for text-based data and information shown for each product. The output of this analysis feeds into a database of individual SKUs available from different brands. We then used advanced search tools to link to an intelligent search engine. This means users today can enjoy the benefits of being able to access and display accurate and up-to-date product information, quickly and efficiently.
Solution
Current state
Cledar continues to support and drive the ongoing evolution of the platform. Using a DevOps approach with 2-week lifecycles, we run 3 environments – development, testing and acceptance, and production (i.e. the live site) – each of which is supported by its own cluster comprising multiple instances to ensure appropriate repeatability. As vendor websites evolve and as the product database grows, we continue to develop the custom software and associated toolset to ensure the Vitaminology platform contains the latest product entries and up-to-date product information.
Cloud-native and container-based
Using a variety of AWS components and containerization technology enables maximum deployment flexibility and scalability to cater to increases in demand, addition of vendor sites, and more.
Advanced data extraction
The solution makes use of OCR and NLP to capture appropriate product information from a variety of different image types, webpages, and files, then aggregates it, removes duplicates and presents it in a standardized, intuitive format.
DevOps and CICD
Using a DevOps approach and Continuous Integration Continuous Deployment, we bring updates, enhancements and fixes to the platform quickly and dependably.
14 APIs
APIs and microservices are developed and used to connect the various different tools, services and data sources (e.g. retailer websites) as part of a single solution.