“The problem we are trying to solve is not really a big one,” says Vinod Vasudevan, CEO of Dubai-headquartered software company Flytxt.
“It is something you saw in the old mom-and-pop stores. They knew their customers personally, they knew when they’d come and they knew what they’d buy. If there was any deviation, they’d know something was wrong or something had changed. This is very useful data. What we’re doing is we’re trying to build intelligence into a machine to help companies make sense of exactly this kind of data – and then make better use of it.”
The product Vasudevan has helped develop to achieve this is NEON-dX, a software solution now on its third iteration that utilises artificial intelligence to process customer data – mainly for the telecom, high-tech media entertainment and financial services sectors – to enable customer service and marketing teams to increase sales and improve customer retention. His ledger of satisfied clients includes Vodafone India, the second largest telco in the world, and African giant MTN, whose subscriber base is more than 230 million across the continent.
What these companies have, he says, is a lot of data that, once analysed, can be used to make strategic decisions about the products and services each individual customer can be offered at every single customer interaction – whether in-store, via the chatbot, the call centre or the app. This is not, Vasudevan insists, simply programmatic ad-serving that reflects recent activity.
What we’re trying to do is build intelligence into a machine to help companies make sense of its data”
“Take a self-driving car,” he says by way of analogy. “It can see an obstacle and can predict that it will hit the object in so many seconds. But it then needs to take the evasive action. It’s only at that point that you can think of replacing a driver. It’s not just about understanding what is in front of you but predicting what will happen with a given set of variables and taking appropriate action.”
In this context, it means looking at past behaviour – usage, billing, product searches – and predicting what a customer might now want and then making an offer to achieve a specific sales goal, whether up-selling or cross-selling, maybe ensuring a person stays with the provider or identifying them as a brand ambassador.
Vinod Vasudevan, CEO of AI-powered software company Flytxt
“In a store, does that mean simply offering them what we know they like? Or is it offering what is best for the store? What product can be tailored that is closest to what they want but still makes sense for the business? Was an offer well received and did it help my bottom line? NEON-dX uses AI to better answer these questions. There is continuous incremental learning in that whole process which is going on day in, day out, with hundreds of millions of customers.
“But just like a self-driving car,” he continues, “AI needs to be able to see the data and take the customer on the right journey to make the optimum purchase. There’s also the possibility that there isn’t a product or service that would work, so maybe this would be flagged and something created. Ultimately, you’re programming a machine not to solve one problem but a whole class of problems – and react to new information. That’s how AI is making a difference.”
What we’ve said is that we want to grow tenfold in the next five years. We’ve been growing at 75 percent for each of the last three years”
The potential returns for customers are tangible enough for Flytxt to offer their clients contracts based on performance. They either sell the software on a service model with a monthly fee or sell on an outright license, but Key Performance Indicators (KPIs) can be built into the deal. The business model is geared towards companies with a customer base of 100,000 to keep a certain percentage of that figure as a control to measure against.
“Our preferred business model is a fixed fee plus success fee concept, which is driven by results,” Vasudevan says. “We know that when it comes to net revenues, we have made an impact of between two and seven percent, depending on which client and the depth and quality of their data. What that means is any investment in the NEON-dX software gets paid back in less than a year, typically something between eight and ten months.”
A new decade
Flytxt has been 11 years in the making. In 2007, after two decades working in various entrepreneurial tech projects in the US, Japan and Singapore, Vasudevan was working in his native India in the telecom sector when he met a quartet of venture capitalists – two Germans, a Briton and a South African – looking for a new project. Together, with Vasudevan at the head, they started Flytxt to develop intelligent marketing software, only for them to pivot in 2010 to machine learning and AI. The result was Neon-DX, whose second and third iterations have become the foundation stones of the company’s current business.
Today, the Flytxt employs in excess of 400 people and has offices in Dubai with R&D departments in India and Amsterdam. With Vodafone and MNT already signed up, backing from shareholders that include the Bertlesman family and the members of the family that founded SAP, and with $50m in new funding recently secured, Vasudevan says there is “plenty of runway” in front of them. They are targeting four geographies: the Middle East and Africa, Europe, Latin America and Asia. Europe is currently their largest market, representing 40 percent of their business.
Omar Bin Sultan Al Olama, UAE Minister of AI
With their recent software upgrade, dubbed Robo-X, they are now looking to expand their offering into other sectors. “We’ve been working with financial services companies on smart wallets and payment solutions and the like for 18 months or so, while we’ve also worked alongside government departments in predicting traffic at complex intersections,” Vasudevan says. “We’ve even done election forecasts – including an analysis for the US presidential election that showed there was a collective shift away from Clinton – which we were able to show just via tweets.
“Our growth targets aren’t in percentage terms of the market, that’s too big and growing too fast to quantify. What we’ve said is that we want to grow tenfold in the next five years. That’s in revenues. We’ve been growing at 75 percent for each of the last three years.”
Liberating not letting go
One of the key issues for AI advocates is understanding concerns about the impact on employment. The fourth industrial revolution has the same potential to reduce headcounts as the first, with computers doing today what machines, engines and looms did in the late 18th century. Vasudevan makes the claim that it will create as many jobs as it costs – and that those jobs will be of greater value.
We want to liberate humans from doing mundane tasks and concentrate on a higher level of work”
“The ultimate goal, of course, is to remove humans from the process, but the reality is that it is some way down the line. Maybe a better way of putting it is we want to liberate humans from doing mundane tasks and concentrate on a higher level of work.
“Look at the printing press,” he says, reaching for another analogy. “It liberated the tasks of copying texts and it helped spread the written word – and by making the book accessible to a wide audience, it created the job of the writer. So, jobs will be lost but others will be created. Why get humans to spend time on something they don’t need to?”