This is an awesome conversation on the cutting edge of language translation.
Note, a large chunk of Unbabel’s customer base comes from the Zendesk PaaS ecosystem.
On the other hand, Unbabel itself would, most likely, create its own PaaS ecosystem to address a broader range of use cases.
Sramana Mitra: Let’s start by introducing our audience to yourself and Unbabel.
Vasco Pedro: I’m the co-founder and CEO of Unbabel. My background is in natural language processing. I grew up fascinated by AI. I could say that has been with me since forever. My mom is now retired, but she was a surveillance linguist.
I did my undergrad at the University of Lisbon. I did a major in Artificial Intelligence with a minor in Comparative Linguistics. After that, I went to Carnegie and did my Master’s and Ph.D. at the Language Technology Institute where I got my natural language processing Ph.D. in 2011.
Since then, I’ve been doing startups supplying different AI technologies mostly within NLP. In 2013, I started Unbabel along with my co-founder João Graca. He had finished his Ph.D. in Machine Translation at UPenn.
We realized that we had this massive overlapping passion for language and the kind of problems that we could solve. One of the things that we were seeing at the time was the huge impact that AI was having in a lot of different fields. We started thinking about what it was going to be within the ability of companies to scale across languages as they grow and enter new markets. That’s a big problem.
Any successful company goes through this where they start going outside of their initial market and need to address several issues related to language. It happens across the entire organization whether it’s about serving the customers, selling the product, marketing the material, and customer acquisition.
All the different functions of the company go through this problem. There is this massive problem that we are trying to solve by what we call language operations. It’s how you run language operations at scale inside an enterprise.
To do that, we started building an underlying translation service that combines artificial intelligence with human translation. It provides the scalability of an AI NIP tool with the quality and robustness of human translation.
We started with the big problems of language operations by applying this to customer service. That was our first use case. It’s our bread and butter. We’ve been focusing on that for the last year. We went through a YCombinator in 2014 and since then we’ve raised four rounds amounting to $91 million in total.
Sramana Mitra: Let’s start in your use cases where you have chosen to apply AI-driven natural language translation capabilities. You are catering to enterprises?
Vasco Pedro: Yes, that is correct. We started within the enterprise case and the different use cases that require language operations. We started within customer services with the fundamental problem of enabling customer service agents to support the pieces of the language. Until Unbabel, the solution was to hire people that speak those languages. This creates a lot of issues because scaling operations becomes complicated. Logistically, it’s a nightmare. You also end up with a team that is across the globe.
What you can do with Unbabel now is you can centralize resources. You can do it from the new short position or auction position. You can take any region and add Unbabel. We integrate with CRM like Zendesk or Salesforce so that agents can do their normal work with it.
Right now, we are focused on non-voice, so it’s email and chat. With email and chat, the agents now communicate with anyone in any language. This has a tremendous impact. The shift from not being language-dependent on your customer service operation has a lot of benefits. One, it’s much more cost-effective.
Typically, our customers see savings of at least 50%. Second, you also have a better service. We see a consistent increase in customer service satisfaction because of several reasons. The primary reason is due to the fact that you are getting much better usage out of your workforce. You also reduce the time for the first response.
Also, you can focus on agents that are product experts. This leads to a much better customer service experience. Using Unbabel enables your agents to communicate with your customers. In doing that, you save a ton of money and you have better customer service. That’s the main use case. Within that, we also focus on email and chat.