- Role of People Analytics in driving revenues and detecting risk
- How AI enriched algorithms can be sensitive to cultural and institutional diversity
- Prospects for U.K. economy and companies post-Brexit
Bloomberg TV: Born to farmers in a small town in Kyrgyzstan, Erkin Adylov has managed to make a career most would only dream of. He worked on his parents’ farm until when he was 16, he won a George Soros grant to study in the US for a year abroad. While his parents would later sell their home to pay for a one-way ticket for him to go to university in Hawaii. A scholarship to the LSE followed before joining Goldman Sachs as an equity research analyst, from there it was co-managing $1.5bn at hedge fund GLG. Then in 2014, he went out on his own with the London-based start-up company, Behavox.
Now the company uses software to monitor employees and grade them on how likely they are to do anything wrong, and some of the biggest investment banks have already been testing it. It’s Surveillance and Erkin Adylov joins us on Surveillance - so it’s almost like a fitting purpose.
Erkin, first of all, you have an amazing life story, which we will have to get back and talk about it a little bit later on. But you are also building a company that is like no other. How did you have an idea of basically giving the tools to a bank or a financial services company to almost monitor every step of their employees but with an algorithm that tells whether they are a bad egg or not.
Erkin Adylov: It’s actually much more than a simple monitoring of employee behaviour. In a way it’s game-changing because what we are actually doing is something called People Analytics. So for the first time, when we come in, we are actually giving banks the opportunity to make money out of their own data. So all of a sudden you can actually make the type of insights out of your data that you couldn’t get before. The insights are then spread into actionable insights or the insights that are relevant for the compliance teams and then commercial insights, the insights that would make your sales teams better.
Bloomberg TV: Erkin, what do you mean by making money? Let’s say you have a trading floor. So, you analyse everything? How many data points for one trader do you have?
Erkin Adylov: So for one trader you actually have tens, tens of thousands of data points.
Bloomberg TV: Such as? How fast you talk on the phone?..
Erkin Adylov: All of phone communications - what you say on the phone and how you say it.
Bloomberg TV: So it picks up words?
Erkin Adylov: It picks out works, picks out your tone and pitch of the voice, so it’s able to classify things like questions, speech rate - are we having a debate or not? Then take emails - you can recognise entities, sentiment; you can recognise places or people that have been mentioned there. And then it gets even more interesting, because then you get into chat. So let’s take Bloomberg chat, Reuters chat, ICE chat - it turns out that there are tens of different chat platforms that traders would use to communicate in the markets.
Bloomberg TV: And it gives you a profile of the type of trader - let’s focus on traders - that you are?
Erkin Adylov: Yeah, it basically gives you a profile. If you are a sales person, it’s gonna give you a profile of all of your relationships, how you are communicating with them, what is the best practice in terms of communication. So it really brings to the fore as to what you have actually been talking about. An example of it would be: a head of sales would be able for the first time see what is actually going on on the sales floor. And let’s say sales floor is 2000 people. So Behavox can actually in real-time summarise every single discussion that is going on on the sales floor and give you a summary of top-ten topics that are being discussed right now.
Bloomberg TV: Where do you draw the line? Are you analysing too much? What’s the line between holistic and intrusion?
Erkin Adylov: That’s a great question because what it really comes down to is who is on the other end of it. Because if the recipient of this information is the artificial intelligence engine or a computer, well I don’t actually have a problem with it. It’s the same thing that happens with Netflix and Amazon - you don’t really have a problem with Amazon watching you as long as there is no person on the other end of it.
At the moment what’s interesting is that banks are already regulated to the extent that they have to record all this information. But unlike us where there is a computer sitting on the other end, there is actually a human being sitting on the other end which is intrusive.
Bloomberg TV: So why is it so important for banks? Again, is it to find the rogue traders? Or is to find out the trading stars?
Erkin Adylov: So it has to be both, right? The way it works is that if you take all of the data, all of the digital footprint of a human being and look at it as individuals, then you can actually apply something called People Analytics - and People Analytics is all about recognising different dimensions of a human being. Humans are not one-dimensional, right? They don’t just come with risks attached to them. They also come with whole bunch of relationships, with the whole bunch of opportunities to make money. It’s essentially allowing you to understand this multi-dimensionality of a human being.
Bloomberg TV: Can you use it to make investment decisions?
Erkin Adylov: For sure. Some of our clients that are in the hedge-fund space are already doing that. They are using Behavox to understand behaviours of their traders and essentially the system helps them to make better decisions. In the testing that we have done, it basically allows traders to achieve 50% higher rate of right decisions.
Bloomberg TV: So what do the banks, or hedge funds actually want out of you? What do they ask you? Are there algorithms or focuses that they want you to enhance?
Erkin Adylov: The key thing is that most of our clients actually start off from the compliance use case. Most of them would start off and say “Can you bring all data in one place because we would like to search it, we would like to be able to interrogate the dataset”. That’s the first task.
The second task is that then they say “Well, we have never been trained by the FBI or we have never been trained by the police, so we have no clue what we are looking for. Can you guys help us?”. And that’s where the Conduct Risk Exchange actually comes into play because with one click they can download every single logic that they need to be looking for or every single model that is actually gonna be helping them understand the dataset that is in front of them.
Bloomberg TV: But like what? Does it say “Francine Lacqua is a little bit too much of a risk-taker" or does it say “Francine Lacqua is gonna do a rogue trade”?
Erkin Adylov: No, it has nothing to do with this kind of analytics. What it does it basically says: here is a bunch of keywords and phrases that historically people said that led them to breach compliance rules. It is always gonna be empirical and it’s always back-tested.
So we will give you a list of keywords, then we will give you a list of behaviours and the list of things that the system should be looking for proactively and flagging them up to the compliance team.
Bloomberg TV: This would be very useful of regulators.
Erkin Adylov: For sure, and we are in fact working with some of the regulators in terms of the POCs and they are trialling the system actively right now. But I think more importantly it’s gonna be more valuable for the financial services institutions because in a lot of the clients that we have it has driven a 99% reduction in the false positives which would then translate into huge efficiency gains for all the financial institutions that we have as our client base.
Bloomberg TV: But how big it can become because if you look at the way CEO or the Board of Directors is ethical or not, could you prevent scandals such as the diesel emission scandal and things like that, or you are limited to financial services.
Erkin Adylov: We are not limiting it to financials services. We are actually engaging some of the corporates out there and given all the scandals that we have seen, certainly from the culture and conduct perspective, this is something that is gonna be relevant for every institution in the world.
Bloomberg TV: How do you scale up from here?
Erkin Adylov: Very easy, actually - hire more people and raise more capital, and that’s actually what we have been doing.
Bloomberg TV: Who are your biggest investors?
Erkin Adylov: At the moment, we have just announced last week that Citigroup is one of our investors but we are also backed by Index Ventures which is one of the Europe’s premier venture capital firms, and the investment from those two firms actually allows us to expand the firm. A lot of the investment that we will be putting into play is gonna be into something called Behavox University which is actually an enablement program. Because the growth has been so phenomenal that we actually need to be training users and training our own people so that’s how we wanna do that.
Bloomberg TV: Alright, Erkin, I have about a million more questions, so I’ll get back to you. Erkin Adylov stays with us, we will continue the conversation with him next but also talk about the outlook for the UK in the post-Brexit world. This is Bloomberg.
Bloomberg TV: Now, let’s get back to Erkin Adylov, the CEO and founder of Behavox. We have just been talking about how data analytics has a power to transform how companies monitor their employees and of course we have been talking about the latest round of fundraising for Behavox. Erkin, thank you so much for ???, I was actually quite taken by how you analyse all these things and how banks can find, I guess, employees that are bad eggs or find star traders. Are we losing the magic? You know sometimes there must be - and I don’t know enough about data and how you crunch data - but there must be a star trader that is different from everyone, that does not fit your algorithms or models. Or is that too romantic?
Erkin Adylov: No, that’s definitely gonna be the case. People are different. Now what’s good about Behavox is that actually when we analyse humans we actually analyse them as individuals, and that’s the power of artificial intelligence and computer systems because you can pump all the data into Behavox and actually ask it to understand human beings in their individuality.
Bloomberg TV: But how often are they right? I was reading up on the art of Da Vinci because of the painting that was sold. Would your algorithms, if you were a trader, and if you were a genius trader which we don’t know, would have spotted him?
Erkin Adylov: Most likely, yes.
Bloomberg TV: OK, that’s how confident you are in your algorithms?
Erkin Adylov: That’s how confident we are in the algorithms.
Bloomberg TV: How can you be sure of that?
Erkin Adylov: We do a lot of back-testing, we do a lot of experimentation. On top of that, there are a lot of datasets that we have that actually allow us to do a lot of testing, a lot of back-testing and make sure that algorithms are fit for purpose.
Bloomberg TV: Erkin, we were talking about your latest round of funding. Are you ever thinking of IPOing or it’s a little bit too soon?
Erkin Adylov: I think over time, as we gain enough mass we will be looking to go public, but I think it’s too early to talk about it right now.
Bloomberg TV: So what needs to happen for you to able… Is there a level that you want to be at in terms of revenue, in terms of sales, in terms of number of clients?
Erkin Adylov: I think it’s more about product-market fit and in terms of what we want to achieve as a company. So clearly becoming a public company is of interest to us but at the same time being a private company actually allows you to do a lot more, it allows you to move a lot quicker. So we will be executing on that strategy first before we actually come to the public markets.
Bloomberg TV: And in terms of your clients, are the US banks more willing or interested in your product? Do you have more reticence from European banks? I don’t know if it’s a cultural difference.
Erkin Adylov: There isn’t really a cultural difference. It’s interesting why they are purchasing, why do they wanna do this. What’s interesting about the banks is that the ROE has been collapsing over the last 10 years, for all the banks, regardless of which bank you take, be it European or the US. And actually what they are coming to us for is saying “Well, can People Analytics transform the way we make money? Can it help us find efficiencies and effectiveness across the varieties of different business lines we have?” - and that’s the main driver of why banks would want to come to us.
Bloomberg TV: We speak to a lot of hedge funds and a lot of them actually invest in artificial intelligence or algorithms that crunch the data surrounds them instead of crunching the data on the person. How is that different? Do they go hand in hand or there are two different ways of looking at data?
Erkin Adylov: It’s actually different ways of looking at the data because ultimately this is very different types of data. Now what’s cool about People Analytics is that your people are very specific to your organization which means that the competitive advantage that you are able to harness using People Analytics is gonna be very specific to your organization and to your people. The type of use cases that we have seen are very different across the different institutions that we have.
Bloomberg TV: Will Brexit hurt your revenues streaming from the UK? Do you think they will they move along with the bankers and the financial institutions?
Erkin Adylov: We don’t believe in any of that so if anything we are actually very bullish about the British economy. I have made Britain my home for the last 12 years and this is where I chose to set up Behavox, and in a way we have no reason to believe that our client-base is going to evaporate. If anything, we have seen all the evidence that the capital, City of London, will continue to be the financial centre of the world.