When looking towards the future we can only dream of where “Neuralink-esque” projects might take human-computer interaction. The reality however, is that our conversations with machines today, whether that be voice or chat, are currently clunky and inefficient. How many of us have frustratingly heard a voice activated call center respond with something like, “Sorry, I did not understand that. Can you repeat?” Or a chatbot conversation that looks like this:
One of the reasons this is the case is because most of the “AI” we interact with as consumers is neither artificial nor intelligent — it is premeditated and pre-programmed for very specific conversations. Pull back the curtain of over 90% of our conversations with machines and you’ll find simple decision-making trees that have very little, if any, “natural language understanding”. Most can only spit out cookie cutter responses based on a specific set of predetermined questions. The heavy lifting required to sift through massive amounts of data and program decision-making trees makes the deployment process expensive, lengthy, and ultimately and unpleasant consumer experience.
This current paradigm needs to change. Data ingestion should be near instant with complex conversations providing value and utility far exceeding the capability of a human operator. Infrastructure for a conversational layer over the web needs to be built. Adding a chatbot or voice activation over a website, app or store should be as simple as building the app and website itself. This is the reason why we decided to lead Airbud’s $4m seed round alongside Michael Neril at Spider Capital and Jon Axelrod at ERA.
Airbud is a platform that allows companies to instantly add true NLU chat or voice capabilities to all their assets, simplifying customer access to relevant information. The Airbud team chose to begin focusing their efforts in three verticals: Healthcare, Travel and Retail. These sectors all share the basic three tenants which Airbud seeks out in a vertical; complex conversations, rich data sets and a transactional outcome. Rich data sets and complex conversations feed each other, and weed out competitors with inferior technological capabilities, while transactional outcomes can prove definitively and quickly that Airbud is driving real tangible business value.
When we met Israel Krush, Rom Cohen and Uri Valevski a little over a year ago, the former two were getting their master’s degrees at Cornell-Tech, and Uri was still a senior developer at Google Duplex. We really liked the team and decided to put them in our category to “actively track”. They had all the right ingredients as a team but were overly focused on voice. At the time it was unclear to us how they were going to commercialize a strong tech stack focused on NLU.
Fast forward 6 months and Weill Cornell Medicine, a leading provider of healthcare in NYC and one of Airbud’s first customers, had an interesting dilemma. Weill Cornell’s leadership was debating a possible lengthy project which could cost hundreds of thousands of dollars in order to build out a chatbot that would enable patients to book an appointment. The other option was to bet on a small team of five ex-8200/ex-Google Israelis who claimed they could ingest their thousands of webpages overnight and deploy a chatbot 10X as effective and flexible. Weill Cornell decided to work with Airbud and have a conversational layer to online booking in production at a fraction of the time and cost as the alternatives.
This was the first demonstrable interaction with a client where the Airbud team could measure speed to deployment and compare it with others during the process. Besides improving Airbud’s product and measuring R&D team metrics like percent of “meaningful and successful conversations” (those conversations where the chatbot could provide an answer to the question asked), the company was able to drive transformational numbers behind a clear business metric — “number of doctors appointments booked”. The company and the hospital will release a case-study soon, but the speed and results of this first deployment got us over the hump in our investment making process.
In looking towards the near future, we firmly believe that every single interaction on the web, whether text or voice, will be powered through a conversational layer — that is why we invested in Airbud. In creating the future of human-machine interaction, we believe the Airbud team has the background, expertise and vision needed to lead the way in making our conversations with machines more human.