Back
When Co-founder and CEO of Verbit.ai, Tom Livne, came to see me for the first time in 2016, he pitched me on their new and advanced speech-to-text engine that he has built together with his two co-founders, Kobi Ben Zvi and Eric Shellef. While we were intrigued by the technology (since it outperformed any other engine out there) we were not sure that any technology in the field is sustainable as a standalone product for the long run, and we were mostly concerned with the business model – how do you monetize it? “This is not the way to go” we concluded with a strong VC determination: “If you have such a great technology, why license it to someone else? Why don’t you become the best transcription agency in the world yourself?” We said good-bye and promised to stay in touch.
About 18 months later, in October of last year, Tom came back with a grin on his face. “So… we did it”, he said. “Did what?” I asked. “Turned into the best transcription agency in the world” he replied. I told him that he always had a good sense of humor. “Hear me out” he asked, and then he laid out what the company has achieved within 18 months: A true human-assisted AI technology enabled service with tens of paying customers and millions of USD in ARR. “I still have most of the money from the previous round in the bank, but really need to move fast – we can disrupt the entire market in a relatively short time”. He then proceeded to outline one of the most innovative and aggressive go-to-market plans I have ever seen. A month later, the deal was closed, the money in the bank and the company is now moving forward at a whopping speed.
At Viola, we have long been advocating to move from horizontal technology platforms into Vertical AI applications for several reasons:
It is very hard to innovate an AI algorithm – most of the innovation is done in academia, it cannot be patented and has been around for a while. Turning the algorithms into generic engines was already done successfully by the large cloud providers, so the bar for startups in that area is really high.
The main differentiation in the technology, therefore, lies not necessarily in the AI algorithms themselves but in the access to unique data that one can train the algorithms on. It is the trained data sets that make the technology unique and create long term moats.
Still, monetizing data is extremely difficult. If the data a company holds is so unique why not utilize it for its own good and create a full end-to-end service? The value proposition is much clearer, it’s easy to monetize, and yes, it requires different skills – not just technology superiority but also operational excellence and thought leadership
The founders of Verbit.ai understood very well that no speech-to-text engine will be able to provide 100% accuracy in the foreseeable future.
When you are aiming to disrupt industries such as education or legal, where it’s literally a “life or death” issue, you really need to provide a flawless, super high-quality service.
Verbit have augmented their engine with a marketplace of transcribers who are constantly correcting errors and perfecting the engine, while at the same time assuring close to 100% accuracy at lightning speeds that no other service in the industry can match.
Tight integration to customer systems that provides seamless “back and forth” of the voice stream allows the customers to completely outsource their transcription jobs and request the desired SLA that meets their requirements.
This is a complex solution as it requires multiple vectors of excellence – technology, go-to-market and operations, and Verbit have a diversified and very experienced management team to make it happen.
Transcription just got a lot smarter, and the academic and legal worlds will gain better accuracy and quality. This is part of the workforce automation which is revolutionizing traditional industries. We at Viola Ventures believe that it is the combination of AI technology and human assistance capabilities, that will allow us to get as close as possible to the 100% solution we are all aiming for.
How will it affect the workforce landscape? Dramatically – from employment models to business models. Stay tuned for more about all of this in my next blog post.
