The main differentiating factor of renovai is that there is currently no solution that knows how to calculate and combine different items based on design knowledge. All existing solutions rely on statistical information. For example, if a thousand people bought a sofa and then bought a rocking chair, then the system would offer the next customer a rocking chair. renovai is the only system capable of understanding design deeply and suggest a design proposal that can convey the right psychological characteristics to each client.
In addition, there are technological solutions that allows automatic classification of furniture items based on statistical analysis of features, but they are limited to classification based on pre-defined features. This method limits the ability of these solutions to combine several items together on the basis of design and shape matching.
renovai’s differentiation is that its unique technology, allows to identify vector properties of a single item, that is, the combination of all the properties of the item. In doing so, renovai makes it possible to produce recommendations that combine several pieces of furniture together, i.e., to match a set of items. For example, a set for the living room that may include a sofa along with an armchair, table and more, which fit to each other from the design and form perspectives.
The differentiating factors can be summarized in several points:
- renovai is the only software that can offer end users an immediate full room personal design, which includes an apartment plan, space dressing and 3D imaging.
- Due to the fact that renovai’s system does not rely on a human factor, the software is able to support an unlimited number of customers and allows renovai to be “Scalable” and robust compared to competitors.
- renovai provides “insights” for furniture companies because the AI machine understands how customers behave, what they like, what their budget is, and their needs, so that the furniture company can more deeply understand its customers. Based on this understanding, the vendors can produce real-time optimization and plan future inventories, and more accurate collection of items, retain customers and increase BRAND LOYALTY in a competitive environment.
- renovai creates a universal design language that allows it to define design spaces and furniture styles in a common and understandable language. This allows the system to catalog and associate the furniture with a variety of design styles and facilitate the design process.
renovai’s system is based on an artificial intelligence engine that combines ML-based algorithms and deep-learning computer vision, with a Semantic Knowledge-Graph, integrated with a Graph Neural Network. renovai’s artificial intelligence engine provides the back-bone for a unique workflow of the recommendation engine, based on three steps:
- Automatic deep tagging – computer vision-based image recognition capability for classifying objects based on multiple properties
- AI engine for matching items – based on multi-attribute labeling, the AI engine evaluates all items that fit the room space, and performs a soft match based on a number of parameters such as design style, shape, material, color, textures and sub-styles, all according to preferences of the user.
- User classification – end users are classified based on previous actions and ongoing interactions with the system, while giving weight to each action performed on the site (wish list, transfer to shopping cart, viewing time, etc.).
The system is able to provide intelligent, visually and stylistically correct recommendations of full scenarios – combination of items, due to its cognitive engine. This engine is a complex knowledge-graph that combines designers’ knowledge with advanced deep-learning computer vision algorithms, allowing the system to provide unprecedentedly accurate design recommendations under each provider’s inventory constraints.
To create a common language between the human experts and the machine, renovai developed a unique design language, partially based on items shape and various other properties, that includes a complete ontology of items, concepts, relationships between concepts, connections between objects, etc. The semantic knowledge-graph contains all the ontological information, information about the items of the various suppliers, the connections between the items, the design concepts and the features of the items and more. The company’s proprietary Integrated semantic design engine supports the need to create interior-designer’s-level quality suggestions, and include combinations of items that match the desired style, colors, patterns, etc. The engine learns by combining the ontology of the “language” of design, design rules formulated by content experts and semantic reasoning that enrich its knowledge about the “design world”. In real time, the system produces arrangements (item combinations) using items existing in each supplier’s cataloged system.