Building a business is tough, something Roy Saadon, CEO and co-founder of AccessFintech, knows all too well. After all, he has founded two of them.
Having sold his first company, Traiana, back in 2016, a number of other opportunities eventually led him to notice a gap in the market with his current venture, AccessFintech: “There was so much innovation and opportunity in the market, it felt like it was time to do the next thing.”
The journey, however, has not been straightforward. “When I first started trying, I knew nothing about the market and had no networks,” he says. “Now, I have scars on my back. I know what mistakes to try and avoid.”
So with the networks now built, the connections established, and the scars to show for it, how is AccessFintech making leaps and bounds in the industry?
That answer may lie in AI.
Just a commodity?
When AI entered mainstream conversation a couple of years ago, the possibilities appeared limitless.
Yet, despite the many conversations, discussions, and debates that have been had in the intervening time, the lasting impact of the technology is still shrouded in a veil of ambiguity. Have then, the impacts of the technology been overestimated? And is the excitement surrounding AI beginning to wane?
“It’s not a question of whether AI is beginning to die,” Saadon begins, “it’s a conversation that has changed due to a more mature environment.”
Having established Synergy, a network powered by data and AI-driven insights, for Saadon, the impact of AI has been much more transformative than other technologies in the sector, such as crypto and blockchain. In these cases, “there was definitely a lot of [initial] excitement.
The transformative nature of it however, hasn’t been as significant as expected. It is now simply part of the infrastructure and technology.”
Rather, Saadon believes that the sudden expansion of the tech has led AI to become commoditised “very quickly”, although he frames this in a positive context. The nature of AI, he explains, means that almost anyone now has the ability to create, run, and develop the infrastructure and models that drive artificial intelligence; models which have been duplicated and used in firms across the industry.
“I think the technology of AI has matured rapidly. The use cases are plentiful, and the conversation is now shifting,” he says. “The questions we have to ask now are, do I have a trusted enough dataset and a wide enough angle on it to allow AI models to train and extract value from it?”
Trusting in the data is a concept that rings familiar for any corporation, large or small, that has employed AI within their operations over the past few years. Skewed, or ‘dirty’ datasets, that have been unknowingly fed into AI models have the capacity to create inaccurate and unreliable outputs, with the potential to undermine an entire organisation.
“AI is really hungry for data,” Saadon reaffirms, “now whether the data is full of false information or not, that’s a different story.”
But it is a story that cannot be overlooked. In a world where data can be regarded as a commodity which can be exploited in the wrong hands, how are firms able to ensure that the ethical implications of data, and the AI that feeds off of it, are mitigated, if not controlled?
When faced with this dilemma, Saddon asks himself two things: “Can I increase the breadth of the contributors of data? And can I use references to make sure that the data is actually verifiable, and identify and extract bad data from it?”
By using these methods, along with critically examining where the data has been derived from and “the continuous retraining of the [AI] tool”, Saadon believes he has a fighting chance at tackling the serious ethical and security concerns attached to data. In his line of work, he is keen to make this point clear: “Protecting data is the number one cardinal rule.”
“So to protect that data, we need to make it a persistent, immovable thing, and give firms the permission to access it. Because by sending data around to every destination, that’s when you lose track of it.” And once data goes missing, he warns, appearing grave for a moment, “there is no way of knowing what people have done with it.”
Optimising AI
When asked about what may lie ahead for the future of AI technology, Saadon reflects for a moment before offering a perhaps more philosophical answer: “Do I want machines to replace every job a human does? Not necessarily, right? That’s not what drives me. But can we do better? Can we optimise?”
Rather than looking at AI solely through the lens of replacing a human workforce, Saadon presents two cases in which AI has helped, not harmed, the firms and clients he works with.
“[With AI] we looked at two things. We looked at jobs that are very repetitive, and as a result, error-prone.” These jobs, such as data clean-up, require rigorous manual checking, where AI, through its ability to clean up data on a mass scale, can help ease the toll the work places on employees.
“The second case is where the volume of data is so large that it is hard to draw conclusions from it,” Saadon continues. “When the overall picture is lost within monumental amounts of data, we can use AI to analyse and draw conclusions from it.”
In these two areas alone, he says, it is easy to see the benefits of the technology, where AI has been employed as a tool to both detect errors and digest data. Although he notes these use cases as two ends of the extreme, they provide a starting point in which the advantages of AI can be clearly measured and seen.
Going beyond this starting point, Saadon envisions the next steps for the technology: “How can AI be used across the whole industry in a way that it does not need separate initiatives for every single firm? What is shareable and mutual, and what can be differentiated?”
How and when this will be implemented across the sector is yet to be known. However, Saadon remains assured that his business, and the technology that it is centered around, will have a lasting impact on the industry.
“When I’m working on AI,” he says confidently, “it is for the whole industry.”
Next interview →
Zema Global
Andrea Remyn Stone
Image: AccessFintech 
