From Keywords to Concepts: Artificial Intelligence in HawkSearch

Lost a sale today on your site? Win more revenue tomorrow with AI-Powered Concept Search

Picture this: you've got every lamp ever made in your brick-and-mortar store and a customer comes in. You ask, “Can I help you find what you’re looking for?” Your customer doesn’t quite know how to describe exactly the lamp they had in mind.  She gives you a vague description of something she saw on her favorite TV show. Luckily you also like that show and know your catalogue well enough to tease out the intention of her description and find just the right lamp for her to buy.  You save the sale that a less experienced salesperson (or one who watches different TV shows) would have lost.  

Now picture the online world. Your website’s search bar is your salesperson. Your visitor is forced to enter some keywords to find the right product. Your keyword-based search can only match the exact words one uses and at best has a thesaurus to recognize synonyms.  That same customer would need to guess at the keywords your web developer entered in your online store, vague descriptions won’t work.  Online retailers lose millions of dollars every day when customers cannot bridge the gap between the concepts in their mind and the keywords used to tag a product.  

HawkSearch's AI has always helped improve the search results around keywords to fill as many gaps as possible. The next evolution of the platform is to leverage AI to automatically identify the concepts of a request through Natural Language Processing (NLP) and image search leveraging powerful, large language models (LLMs) to identify the concepts behind the search. This innovation will find the conceptual match to connect your visitors to the products they want to buy.

The Search is Over: Your Secret to New Sales is Here

At HawkSearch, we’ve pioneered different technologies for you to grow sales through your best virtual salesperson: your search bar. Our “AI Multipliers” drive more relevant search results by integrating the latest advancements in AI LLMs and Vector Databases, to give your customers the ability to use conceptual phrases, in any language, or even pictures in their searches to find the products they want to buy. Let’s look at an example:

Conceptual Search: Beyond Keywords

A Modern Day Hat Chase

Imagine - your daughter idolizes Sophie Turner, the Queen of the North. All she wants for Christmas is a beanie, one like what Sophie wore. But you're lost, “Where on earth can I find this hat? What's that beanie even called?!”  

With HawkSearch's conceptual prowess, your quest is simplified. Perhaps you're from Canada and you input "toque" instead of beanie on a US-centric site.  HawkSearch recognizes and translates your regional jargon, ensuring that you find Sophie’s coveted beanie. Or maybe English isn’t your first language, and you end up typing “Sophie Turner’s gorro de invierno.” The concept of your search is the same, regardless of the language, so AI Multiplier marches forward without skipping a beat. Better yet, you can drop the image of Sophie or a web link into the search bar and ask for a hat like the one in the picture. Instantaneously, you're directed to that perfect beanie. Christmas morning promises an ecstatic daughter.

‍

‍

The keywords, even the language, are not what matters – it’s the concept that matters. The concept is the shape of the hat, or the fact that Sophie Turner is wearing the hat, or what the hat looks like, or all the above. In a physical store your customer relies upon the concept, possibly even a picture, to work with your salesperson and the depth of this human-like interaction is what turns browsing into buying.

Behind the Scenes

HawkSearch leverages the power of LLMs trained from trillions of documents to accumulate general knowledge with a Neural Network. The LLMs can be used to create Vector Embeddings which are functions that convert text, images, and even video to return vectors of hundreds of dimensions that numerically represent the concept(s) behind the content.

To support this concept search we add pipelines to your existing product catalog, brochures, spec sheets, and even third-party reviews to create vectors that represent a neural network of your information. Each time your customer performs a search, we run the same vector process again to match the request to the neural network in our system. This matches the concept of the request to your data on our platform.

Moving Forward

The innovations in conceptual search at HawkSearch are intended to extend the current capabilities our customers already enjoy – not replace them. The concept capabilities can be blended with the keyword searches for example. They can also be presented as an option like “still not finding what you’re looking for?” The opportunities are endless, and we are excited to work on them in partnership with our customers. More to come!