Outsell for Startups podcast Q&A: Ruth Pickering of Yewno

‍

In 2016, AI companies managed $5 billion in funding across 658 deals. That’s a 62% increase from 2015. That’s a lot of benjamins, pounds, bitcoin, or whatever currency you’re throwing at this space.

Investment comes from its promise. Already, powerful AI applications are letting companies move at a much more efficient pace. That said, AI is still very new, and the talent pool that understands how to utilize it is limited. There are even fewer who understand how to deploy the logic of neural networking, cognitive science, and computational linguistics. One company that’s figured out how to harness that talent is Yewno.

Listen to the episode now on iTunes, Stitcher, or Sound Cloud.

I interviewed Ruth Pickering, Yewno’s co-founder and chief business development and strategy officer. Her experience spans from media technology and telecommunications to AI, where prior to Yewno, she served as managing director of BT Wholesale, was one of the Global Telecoms Business top forty-under-forty in 2010, and then was shortlisted for woman of the year in technology in 2011.

Yewno is helping people uncover the undiscovered through its new inference engine, which introduces an entirely new approach to knowledge discovery. It essentially mimics the human brain and the Yewno inference engine incorporates machine learning, cognitive science, neural networking, and computational linguistics into a highly visual solution to enhance discovery and human understanding. So for example let’s say you’re a researcher for a biomedical company working on cancer research. You have to navigate through thousands, if not millions of pieces of data and content to find the research that you need to do any kind of experimentation or even make progress with your own research. Utilizing traditional search can make this quite complicated, but what if you had the power of AI and neural networking to actually help you identify these resources at an accelerated rate and as result, develop a cure faster.

Yewno has done an amazing job in terms of getting out there, becoming known, causing buzz, and as a result we believe the company is on acquisition lists for a number of publishers. 2018 could be a bit of a tricky year as the company needs to cement its position and differentiation in an effort not to become overwhelmed as AI gets bigger and more openly developed and accessed.

Key metrics: Founded 2011. Number of employees 38.

‍

‍

‍

You have always been searching for new things and new disruptive technologies. Why AI and neural networks? What drove your interest specifically in those technologies?

Ruth: It’s really interesting. Most of my career was in Europe, but I did a lot of global roles. I started off in product development. Within product development you’re always trying something new, something that doesn’t exist. In that space, what I really liked was the fact that we would come up with ideas. We would work really really closely with our customers, with our salespeople and we would test things. Actually the products that you come out with at the end are often really different to what you originally envisioned having.

As my career progressed, I was running a super fast broadband program for BT. When we started that program (it was a one and a half billion dollar program) people said to us, “Nobody’s going to want this. Nobody needs that much bandwidth.” But we all knew it was coming, we all knew the applications coming down the tube, but your average man in the street maybe didn’t see those things happening. You kind of have to work your way through all these things, talking to different people and understanding. For me, tackling AI having come from super fast broadband, and knowing what fiber could do was very interesting.

When you start looking into it, and you consider how search has changed our lives in the past 20, 25 years, young people can’t even remember what it was like to live without search. Then when you think about what AI can deliver, we’re just really at the very very beginning, it is definitely, for me at least, the most exciting place in the market to be.

So let’s just take a minute and just talk about AI, before we even get to you Yewno because there are a lot of exciting things happening with AI, and we’ve really, I feel like, just scratched the surface of what our capabilities are with AI. Could you talk a little bit about where you see AI going, maybe in the next one to five years, and maybe how Yewno is going to be applying it?

Ruth: Yeah, so for me, I think sometimes people have a bit of a kind of a shock reaction of AI and are think about some negative connotations because a lot is published in that space. I always think about it, and we actually sometimes in Yewno talk about it, is augmented intelligence, rather than artificial intelligence. It’s doing all the things that you or I could do, but on a much much bigger scale. I often think about it on a personal level if you’re a researcher, if you could read 10,000 articles rather than 20 articles, how much better would things be.

When I’m talking about AI and representing AI for Yewno sometimes people think AI is going put them out of a job and I do not think that at all. I think AI is going to make your job so much more interesting. A machine can do huge amounts of high volume repetitive work, but will save the person that time, and then it will give you this information, which is so much richer, which is so much deeper, that you can make much better decisions and you can do the more interesting parts of your job, but it’s not about taking away jobs.

What do we say to all these people that have a fear of losing their jobs, like you just said, to the robots or the machines or the new AI technologies that are developing. I feel that some people are very skeptical and very non-optimistic about the technology and have a fear that their going to lose their jobs. From a service level perspective, as a researcher, this isn’t a concern to your point because it’s really optimizing the way you garner information and use it. What would you say to the people that do have a fear of robots, AI and technology about their jobs and their future?

‍

‍

Ruth: Yeah, I think it’s fair to have that concern because I think there’s been a lot of hype in the press, and when you hear about AI, you don’t know specifically what it’s going do. You could imagine it doing things like taking away jobs but in the realms that we see it being used, it’s providing this ability to synthesize what a human can do on this enormously different scale, but you still need a person to make a decision. There are elements of AI, if you take one extreme, where you could potentially see a machine making a decision. In terms of the implementation that we’re doing here, and what the work we do at Yewno, we’re not actually doing that decision making.

So for example if I take the research applications that we have, we’re proving a researcher with this incredible richness of information that they could never find themselves because they could never read tens of thousands or millions of items. We’re not telling them the answer, and if you take what we do in the financial sector, again we can assimilate this enormous volumes of information that no hedge fund manager, no analyst could ever ever assimilate themselves. But we’re not telling them what to do with their money. I think it’s the fact that you can actually help somebody do their job by providing them with much more accurate, unbiased, reliable information, but you’re not actually doing the elements that I would consider the human parts.

We’ve really just by talking about AI circled around what Yewno is. I want to dive into Yewno, but I want to start with it from the problem. What was the problem that was identified that Yewno was going to be built to solve?

Ruth: I think the problem that we noticed was the enormous volume of information that exists already and that is increasing every day. It’s not only that the volume of information is increasing, but that it’s very fragmented across huge ranges of different publishers, different websites, and organizations. Some is internal, some is confidential, and some is open-access. We see there’s a big difference between information and knowledge. I know that personally, you hear a statistic and then you think, “Oh I want to find that. That was a great statistic, I wan to use it.” And then you spend ages trying to search for it, you could spend a really long time trying to find an information source. The problem we’re trying to deal with is take away this issue around the volume and fragmentation of information and to actually help people find what they’re looking for and acquire knowledge, which is a very different thing from just having realms and realms of information at your fingertips.

So if I’m understanding this all correctly, outside of Yewno there are a lot of search solutions to help filter through numerous data silos, content silos and bring that all together but would you say the core of Yewno is that we employ machine learning AI and basically think of it from a neural networking standpoint to really find the best information an accelerated rate. Is that what it does?

Ruth: Yeah. So when we talk about AI or augmented intelligence, what we’re trying to do with the neural network model that we’ve created in Yewno is to replicate the way the human mind works, and there huge benefits around that. Essentially, in terms of the way our algorithms work, we are able to ingest these huge volumes of information and also across many different languages, many different types of data, formats of data, and then we’re trying to process it in the way that a human brain works, and a human brain would create inferences and links. But we’re able to do that on a different scale. That, to us, is really the uniqueness of the product in any sense whether you’re looking at a financial product, whether you’re looking at a biomedical product, whether you’re looking at a research product, it’s that inference creation that is at the essence of what we do.

What does the team that handles neural networking and AI applications look like at Yewno? I always have the question in my head, “What does that look like to employ a technology like that, when you break down a neutral network like that?”

Ruth: So we have a big data science and an engineering team, we have a content team, and I think that it’s probably the largest percentage of people employed at Yewno. Within that, they would break themselves down to different kinds of roles and what they’re all looking to do is absolutely fascinating. So we have somebody who got a PHD in applied physics from Stanford. You get lots of people with PHD’s in computer science as well. They’re all taking different elements of the neural network model and breaking it down and understanding and testing things and some of the research takes years. It’s not something that you do and a week later you have some variation. People have been doing research for a really really long time.

One example is that we started with English only products. We always knew we wanted to do other languages both independently, so if you want to research in German or search German financial information, you can. But what if want both German and English? We knew that we wanted to create, for example, a multi-lingual, knowledge space. The work to do that was a huge project and we had a team of people focused on just that for months or possibly years in terms of recreating and making sure that we could create this multi-lingual environment, which preserved the semantic similarities between concepts across languages.

With Yewno I noticed that you really serve four different industries. Are the products to serve these industries somewhat different?

Ruth: That’s a great question. I think there are really two major components in terms of the foundations of our company. First of all, is the data science part, and the algorithms, and the second part is our content set. What you find in terms of the way our technology works is that we’re ingesting these huge amounts of information indifferent formats, then we’re running a series of algorithms and we’re creating this enormous graph network. The graph is a fantastically flexible mathematical instrument you can repurpose it in many ways.

If, for example, you’re in the financial sector you probably don’t want to find medical device history articles. We will take the relevant information sources for that sector and you’ll look at the financial products and they’ll look very different. You’re not going to see a graph visualization, you might see an index in front of you. You might see some text that’s been written by a machine but actually underlying it, a lot of the technology is reused, a lot of the content is reused. For example, we use patents for our financial sector, we use patents for the academic sector as they are interesting to both. What we try and do is reuse the technology and reuse the relevant part of the content set and then we’ll create a unique visualization that’s appropriate to that vertical.

Great, and if I recall correctly, the industries you’re currently serving are publishing, finance, biomedical, and education correct?

Ruth: Yes, and our thinking was that we started off with education because education covers every single subject area. If you look at the market today, there are a lot of people who focus human capital on sciences, a lot of people who do a unique element and we wanted to make sure that we built our company and we built our technology in a way that was interdisciplinary so that we knew we could work across all different types of information sources. Starting with education is brilliant because you need that breadth. It was probably a difficult place to start but it meant that having started there we could reuse, for example, a lot of the content that we built for other verticals and also, biomedical you describe it as a different vertical and it is, but it is also used in education, so medical schools will use the product, for example, in the same way that a pharma researcher could also use the product.

Just going off the biomedical example, perhaps it would be helpful for our audience to hear a use case about how a client has used Yewno to help them succeed and find the information that they’re looking for.

In the biomedical space we have a great case study on somebody who is looking to repurpose existing drug molecules to cure rare diseases. I think in that space, there’s a huge amount of money and it’s spent largely on the diseases that will impact large numbers of people. Things like cancer. In this particular research, based in Germany, they had a real problem at hand. It was an outbreak that happened. It was only this June and they needed to come up with a solution quickly so they pulled together a virtual team working across the world. They used the Biomedical product to find mechanisms of action that could be specifically used at that moment in time. Existing drug molecules, they didn’t have time to invent something new, they didn’t have ten years to go through the drug development pipeline. That’s a brilliant example of how you can use the platform and the contents that it contains to help solve a real world problem.

So if I’m a researcher and I want to get access to it, is it a subscription based model?

Today, the way in which we offer the product is subscription based, so we provide it to universities and libraries, but we are looking, in the future to a model where an individual could subscribe and potentially, some kind of public offer.

What does the competitive landscape look like for Yewno?

That’s a really great question. I think in terms of what we see in the market, a lot of people have said to me, “Oh, I’ve seen a product like this and it had a bit of a graph.” And there are definitely some smaller graph versions that have been manually created historically. We don’t see anything the same as our product at all in the education space. There are some variations of the theme in finance, but what a lot of people I’ve talked to are really in the NLP space rather than the neural network space. I like to think that we’re ahead, but we’re sitting in Silicon Valley. There could be 50 people sitting at home doing something amazing that they’re going launch tomorrow. So we’re always on our toes that’s where we invested heavily in the data science side because we like to think that we’re ahead now and as far as we know we are, but we want to make sure that we are investing in our technology to maintain that position.

How big is Yewno today?

We’ve got about 38 people. We have three people in London, nine people in New York, and the rest of their team is based here in Silicon Valley. We’ve got about thirty customers using our education platform at the moment we’ll be making some press announcements about some recent finance customers wins that we’ve signed in the next couple of weeks.

Can you talk about your growth?

We launched our first product in June 2016, so just over a year ago. We signed up a handful of customers in the first six months. So I guess it’s kind of a sextoupling of our customer volumes. In the next year, we recently made announcements about our financial intentions and financial vertical and we’ll be announcing some customers in the next couple of weeks, so that’s pretty fast moving as well.

Where would you say Yewno sits in the startup lifecycle? Are you looking for funding in the future?

We completed our series A in November 2016 and we are embarking on a series B. Our initial investors are staying with us and we’re looking to add a couple of new investors. I wouldn’t project much further than this but I wouldn’t say that fundraising after series B is out of the question. I think we’re kind of out of what I will call the early startup phase. I think when we hit 100,000,000 items of content, we felt that we were in the big pool if you know what I mean. We’re still early stage.

What would you say is the biggest challenge Yewno faces?

I think what I find, and you asked a little bit about AI at the beginning, I think that there are some areas where people are slightly suspicious of AI, and I think we’re one of the first companies that have come out with a real-world AI product that you could have on your desktop today. I think sometimes when you’re talking about AI there’s still a slight fear of it. I think when people actually see the product in action and they understand how they work, then people are very comfortable with that, but I do think there’s possibly a slightly negative reaction to AI in the big sense of the word. That’s possibly the biggest challenge you face.

Is education around AI part of the selling process? Is really identifying your potential clients and educating that this AI solution is very powerful, is that part of it?

Yeah, so I think when we meet with customers they’re all different. Some people know a bit about AI from their previous history. Some people have teams in their organization who deal in AI a little bit. I think a lot of what we’re trying to do is to explain, we’re a new company we’re not a global brand you always have to explain a bit about your organization. We like to explain how our technology works because it is different and unique. Then I think generally speaking, we have a very positive reception and people are pretty excited. I think once people understand what you’re doing, we get this huge interest and excitement that people want to hear more, and they want to understand how it can help them.

Where do you see Yewno in one year?

We’re continuing to grow. We will continue to grow our office here in Redwood City. We’ll be continuing to grow our office in New York in particular, and I think we’ve got a pretty busy product development calendar. We’ve got some exciting things coming up in the new year in terms of enhancements to our existing portfolio and some new things that we’re working on. I think we’ll still be in the for vertical you mentioned, so we’ll still be very active in education and publishing. We’ll be growing significantly in finance and also in the biomedical space.

Where do you see Yewno in the next five years?

So the next five years I’d like to see us as a more established brand. I’d like to think we’ve got our public product out there, and so not only institutions, not only investment houses, pharma, but anybody can use Yewno in their home if you’re a lifelong learner in any situation. That’s where I see ourselves transitioning from the very specific verticals with very clear target customer basis, to a product that is accessible to everybody.

Yewno was awarded Outsell’s Emerging Company of the Year for 2017. 100+ CEO’s in the industry voted for your company. Why do you think they voted for Yewno?

‍

‍

Well, it was pretty stiff competition that afternoon, and because Yewno starts with a Y, we went last, and so I’d watched all of these other finalists and they were great companies. I was really impressed. We were delighted to win in the face of such stiff competition and I think the only thing I can say really, because people were doing very different things in this space, the one thing that did strike me is that there weren’t many people who were doing something as interdisciplinary. I’m not sure if that’s the angle that people liked, the fact that it was really across all different types of vertical and types of information. It was a huge compliment to be recognized by Outsell, and by the rest of the industry because as I said, the other companies there were pretty stiff competition for us.

What would you say has been your greatest learning so far, or the company’s greatest learning so far?

I think we were amazed by the customer reactions. Once you get to talk to the right person and you explain what you’re doing the level of excitement and the reactions that I’ve had in meetings are reactions I’ve not had from customers in meetings for my entire career of 20 years. That sort of thing has been fantastic. I think for the whole company, it’s been an amazing journey in terms of one minute there’s two of you in the office, the next minute, there’s four of you in the office then there’s eight of you in the office then there’s sixteen, and to go from two people to 38 people in just over two and a half years is quite rapid growth, and I think the other thing that has surprised us is that we originally were planning to roll out in the US and the UK. We had our English language product, and without doing any marketing or advertising or PR at all, we’ve ended up with queries from around the world. We have customers in six countries. We did not expect that global demand for our products. I think it was pretty much word of mouth so that’s a great compliment, really. That’s something we’re really happy about but that’s probably the biggest surprise.

Thank you for reading this Q&A interview with Ruth Pickering. This write-up is only a portion of the full interview. To hear the full interview, please listen to our podcast “Outsell for Startups”. It can be found on iTunes, Stitcher and Sound Cloud. You can also find more information on Outsell For Startups at our website. Thanks for reading!