Inversys, Banco Santander… We’ve seen plans to personalize financial advice with AI, where do you think this market will go?
Counseling aims to evolve towards an increasingly personalized counseling model. This personalized consulting model would, logically, require a large amount of costs combined with the costs of dealing with standards and regulations. This makes consulting models unsustainable at some point from an economic point of view. It is precisely technology and especially artificial intelligence that will allow us to meet those challenges we face within advisory services, including disruptive technology that will allow us to create more personalized advisory models by leveraging enormous amounts on the one hand. This is unmanageable data unless AI or some variant of it is used. On the other hand, it will allow us to develop models from an economically sustainable point of view, in terms of costs. All this, logically, with the need to comply with the regulator.
Therefore, these two variables, increasingly personalized advice and compliance with the regulator, technology and more specifically artificial intelligence, a key element to achieve them.
In an increasingly regulated world and where commissions for advice are more openly asked, what role does AI play and will it play and where is the advice industry going? Do you think it will meet the demand of investors who are not advised because of the high costs, as has happened in the United Kingdom and other countries?
It may be but I think AI will help reduce overall consulting costs. Granted, in some sense this cost savings will allow consulting commissions more accessible to that client profile. One of the added values of AI, as a process improver or technology, would be to reduce the costs of the consulting service development process and, therefore, allow for more competitive pricing in some cases. Allow for a single consultation model that is more profitable and, in others, allows access to consultations for people who cannot afford the high fees.
Openbank and Openfinance (Inversis) are two great success stories of AI applied in finance. What kind of products do you advise with Node.AI?
We have two different solutions we are developing with OpenFinance, in one case, CoPilot and OpenBank. The first of them is the “Intelligent Advisor”, whose models are powered by artificial intelligence machines to pick and choose investment alternatives, guaranteeing the selection of the best alternative based on parameters defined jointly with clients.
We use multiple layers of filtering, which we call regularity, another layer called connectivity, and another called consistency. Regulatory and Affinity Layers are both natural and normative layers where the choice of product or chosen alternative is adapted to the client’s risk profile according to the nature of the mandate held by the client. All the parameters are adjusted in the sense of the client’s profile, so it is compatible. But the most important thing for us is the outlier and it’s another thing than putting another layer through the intelligence engines where, with a certain level of probability, the execution choice in the portfolio is currently the right choice. And for the next two or three months. Through our profitability evaluators and risk evaluators for each of the products we analyze, we guarantee that this portfolio will not have to be rebalanced within a few days because we have introduced an asset that will disrupt us, not now, yes. The characteristics of the portfolio may be within days.
Within this solution, we have developed an explanation of the selection in the tool, organized in such a way that it speaks to the financial advisor through generative AI, justifying the reason for the selection of that asset. This is very advantageous for companies as you can guarantee the same dialogue by all potential salespeople when talking to customers.
As for the solution we developed for Open Bank, for its EnergIA broker it is nothing more than profit estimates calculated for different windows – one month, three months, six months and twelve months – for the time frames of the US market and the US market. European market, especially S&P and Stoxx .600. Here everything is created by artificial intelligence, by our own models, where there are about 55-60% accuracy levels between the estimates we make and what actually happens, which leads to significantly higher levels. Accuracy. In fact, the CFA Institute states that analyst accuracy is 35 to 45% accurate. Therefore, we are trying to leverage artificial intelligence to provide us with process-optimizing solutions in the world of asset management and web management.
Not in vain, we are not technologists implementing technology, in this case, artificial intelligence in the world of asset management or web management. In our company, we are the majority in the asset management world, we are ex-UPS from Switzerland, France and Spain and we are from the asset management world who went to technology to find the solutions we enable. We improve the processes we already know. This is important to the extent that we have a critical sensitivity to know what our models are generating as output and to ensure that, in fact, the output generated is not random. The reason behind it. This is a very important feature that is very difficult to implement in an organization and business model, if it does not come from asset management or web management, it has been sensitive to management and investment management for years.
Can a machine be taught to manage the uncertainty of the market and global society?
Machines cannot be taught to manage crises. In fact, machines must learn from and recover from events that occur. Without going further, we have established a rule that in case of any crisis – a black swan, which is not thought about, like an epidemic, our models will automatically stop generating estimates. They stop until a few days or two weeks have passed and the crisis can be introduced or incorporated into the models so that they can begin to evaluate and learn. So, obviously, AI helps us a lot in most contexts, but it’s true that when focusing or facing a new situation of uncertainty or crisis, waiting and restarting, stopping and restarting the machines is the most sensible thing to do. In a few days, when the models start feeding data from the crisis.
Although this is something that has been going on for years, we are still in the early stages of AI development. At what point is the control?
Regulation must go hand in hand with the growth of innovation. If it goes forward, if it goes backward, it could leave “particularly large holes” that could pose irreversible risks to the customer, the investor. Protected by the regulator.
In this sense, I believe that regulation should go hand in hand with growth and innovation. Of course there have to be controls, because you can’t control or eliminate impact or innovation or artificial intelligence, but you have to continue to protect the individual and the small investor.
There is the improbable level of artificial intelligence, which is more difficult to control and is closer to deep learning, or the artificial intelligence level, which is more complex to interpret and has more deep thinking steps. Learning models allow us to extract the reasons that produce these estimates. Regulation has to move forward and it has to go hand in hand with both financial institutions and technology companies because it is something that is here to stay. And there should be an adaptation on the part of the regulator to regulate the activity of the innovation and on the part of the technology companies to try to provide their customers with the maximum information that can guarantee that transparency. Ultimately it has to be a joint effort of all the actors in the field and, as in other cases, we have to walk little by little to discover new terrain as we move forward in the face of innovation and innovation. .
