Technology was only used to publicly share our messages on social media. But Karan Walia, Cofounder and CEO of Toronto-based startup Cluep, wanted technology to understand what you publicly shared and your emotions. “We think we make choices on logic alone but that’s not really the case. Our feelings and connection to things happen on a deeper level in the subconscious mind – and this influences most if not all of our behaviors and decisions. The realization that it’s not our logic but our emotions that drive most of our choices is a powerful insight for brands who are looking to better understand their consumers’ psyche.”
An exciting player in Canada’s tech landscape, Cluep harnesses AI to give marketers the ability to serve mobile advertising to people based on content publicly shared on social platforms. Cluep was founded in 2012, the result of Karan’s obsession with AI and quest to build the ultimate digital personal assistant. “I teamed up with my younger brother Sobi Walia and teen programming whiz Anton Mamonov to write a novel AI algorithm that can suss out emotions behind millions of words and phrases. My teenage partners and I ended up using this algorithm to start a company called Cluep and let brands serve ads to people based on what they publicly say and how they feel on social media.”
Describing how their novel algorithm works, Karan shares, “We wanted our AI algorithm to understand things like slang and the fact that words could be used in multiple contexts. For example, if someone says, ‘The new Apple iPhone is sick!’, is the person feeling ‘excited’ or ‘sad’ and did they mean the fruit or the device? Using deep neural network architectures and recurrent neural networks (RNNs) to help perform word and character level-based learning, we were able to learn ‘deep’ representations of the grammatical structure and the semantics of a sentence. In my example, the addition of ‘the’, ‘new’ and ‘Apple’ hardly change the original feeling associated with ‘iPhone’. But it completely changes when the phrase ‘is sick!’ is included alongside it. RNNs are good for managing these kinds of cases, which are frequent in natural language. For instance, the best way to know that ‘Apple’ is referring to the company and not the fruit is by seeing the word ‘iPhone’ beside it. This can only be determined if the words surrounding ‘Apple’ are taken into consideration in the neural network. With these two approaches combined, we were able to get an accurate way to parse and classify text into one of the 8 different feelings — Happy, Loved, Excited, Hopeful, Scared, Sad, Horrible and Angry.”
Unsurprisingly, advertisers around the world were ecstatic to use the algorithm and spot opportunities with existing and new customers. “Cluep’s patented AI technology is now used by thousands of brands which include Amazon, Microsoft, Spotify, BMW, Ford, Mercedes-Benz, Red Bull, Starbucks, Coca-Cola and McDonald’s. Our clients are seeing conversions and click-through rates that are 5x-10x times better than industry standards. AI is disrupting the traditional model of advertising and how we think about targeting and connecting with people – it’s actually making it more human and emotional,” Karan insists.