We asked our contributor pool. Here’s a selection of responses.
Stephen Findlay, CEO, BondMason
From the perspective of the investor, there is likely to be little distinction between AI products and classical fund managers when it comes to making investment decisions and deploying capital. The two key questions will remain the same – what return am I getting in exchange for risks, volatility and illiquidity of the service; and how well can I trust the counter-party. AI may be able to deliver higher alpha – time will tell – but non-AI strategies are likely to survive alongside.
Nonetheless, AI strategies are well positioned to grow, continuing the longer- term shift to trackers and more passive fund managers – our service at BondMason achieves exactly that for clients looking to allocate capital to P2P Lending on a more passive basis – which is why we have seen one of the highest growth rates amongst P2P Lending platforms in 2016.
Dean Young, eWise
As Artificial Intelligence merges human insight with automated analysis, AI can help financial advisors, bankers, fintechs and their customers predict cash- flows and create disruptive API based solutions.
Today financial institutions and fintechs are collaborating to create a broader ecosystem where value added services are quickly implemented through the use of APIs. Not only will banks provide open API to access user’s payment accounts post PSD2, but Fintechs will leverage the data in order to contribute to the global value creation chain.Many initiatives in this regard will require Artificial Intelligence. An example is the eWise transaction categorisation API called CaaS. This Categorisation-as-a-Service API automatically categorises user’s financial transactions; augmenting crowd sourced data with AI. Financial institutions and Fintechs can connect to the CaaS API and develop their financial app based on high-quality transaction data categorisation.
Through AI used in financial transaction auto- categorisation, businesses can understand how users interpret their spending, and predict customer behaviour. Transaction categorisation will pave the way for understanding users’ spending patterns and behaviours, helping developers to launch better Money Management tools enabling cash flow, forecasting, spend analysis, auto-budgeting and financial calendars. These tools are mostly used by consumers to track their financial health.
Another use case for transaction categorisation is to facilitate the highly accurate credit analysis needed to enable loans for individuals and businesses. Credit processing is speeded up by accessing and authenticating user’s transactional data, as smarter income verification and spend analysis will reduce the risk of default due to a deeper understanding of a borrowers’ financial history. AI led financial categorisation can also be used for fraud detection. Mining data from user’s transactions can determine normal and abnormal spending behaviour, giving invaluable early warning to firms of all sizes.
The CaaS platform works by processing all of a user’s historical transactions from their bank, credit card provider, brokerage firm, pension, utility provider or loyalty programme. Regardless of the language in which the data is processed, the categorisation API understands it. CaaS includes the “Truth API”, which is the Artificial Intelligence driven, scalable approach for transaction auto-categorisation. The self-learning platform is responsive to local nuances, enabling business to understand their customers spending and focus on their business strategies.
Yet, auto-categorisation can never be 100% accurate using a fully automated AI. Machine learning has a degree of error, and spending descriptions can be subjective. CaaS negates these irregularities through “Assisted Categorisation,” which allows users to control their preferences and work in collaboration with the Truth auto-categorisation API.
CaaS therefore fluidly manages categorisation; coverage, category calibration and custom categories. It synthesis human and artificial intelligence in order to fully answer user’s expectations and get a better user experience through the categorisation process.
Financial services and Fintech firms will continue to tap the economic and engagement value of enriched data, and CaaS architecture is an example of the intelligence technology that can be easily implemented through an API. Once companies can source (aggregate) standardise tag, and enhance data data, a structure can be built based on their clients’ needs, ushering in era of intuitive and predictive customer service.
Jenna Cheng, Envestnet | Yodlee
How Financial Institutions Can Boost Engagement With Chatbots
When it comes to the fintech world, bots have already found their way into a few basic functions with apps and some forward- thinking financial institutions, but they have a lot more potential. They can help your customers with basic information and tasks, such as opening an account, making mortgage payments, and even funding an ATM.
But there are downsides as well, and fintech professionals and bot developers would do well to consider them. Will the bots be too impersonal and turn customers off? What if a bot misunderstands a question and processes a larger payment than was requested? How easy is it to hack into a bot, or trick one after identity theft?
Bots are here to stay, and are going to grow in their use and acceptance in the fintech world, and many others.
Chatbots are not only a money-saving automation tool, they can automate simple tasks, such as opening a new account or transferring money from one account to another. Better still, there’s no need to create a separate chatbot app; these can run on your existing websites, so they’re readily available when customers visit your site for assistance and information.
Of course, one of the challenges with a conversational/chat interface is that a customer may ask questions that break the system. If you ever played with the Eliza computer therapist game, you know how easy it could be to flummox that program. While today’s chatbots are/will be light years ahead of Eliza, there are still ways to break the bot and necessitate a reboot.
This means you would have to start with specific tasks it could do, such as processing an account application, or completing a bill payment. This also means the bot needs to have a wider understanding of language, since people use different words to describe the same process. (And it will be even harder, once we move to a voice-activated chatbot, and the system tries to understand different voices, pitches, and even accents.)
There are also a range of exciting innovations companies like Envestnet | Yodlee are integrating into their environments to further power the bot experience, based on data analytics, artificial intelligence, and machine learning.
Imagine a consumer opens up her banking app, and is interested in buying a $200 pair of shoes. This scenario illustrates how actionable data and insights are used through descriptive analytics, predictive analytics, and prescriptive analytics.
Descriptive: Our consumer asks, “What’s my checking account balance?” She does the mental math to figure out if she can afford the shoes, or if it will hurt her immediate financial plans, and makes her decision based on this simple question and answer.
Predictive: “Can I afford to spend $200 today?” The bot, with the internal financial data, creates a forecast based on historical recurring income and financial obligations. The system can look ahead to the next couple weeks, and let her know what impact the $200 purchase will have. Predictive analytics is one area that Envestnet | Yodlee is putting a lot of work into.
Prescriptive: “Should I buy these $200 shoes?” This is where the bot acts as a financial coach. At this point, the bot will understand insights about the consumer and her behavior, people just like her, and the long-term financial goals she’s trying to achieve.
The bot will tell our consumer, “that will impact your budget for the month, would you like me to give you a couple options on how you can save money?” The bot could even see our consumer’s location, and could find where the shoes are being sold for less, or even locate a coupon to help reduce the price and make the purchase less impactful on her financial situation.
Financial institutions that want to boost their engagement with chatbots need to build or partner with vendors who can develop the machine learning-based systems that will continuously analyze data that creates insights that drive personalized financial conversations with their end users.
Lucas Montano, CEO, Planejei
In 2015 we had a major discovery: personal finances is not about numbers and charts, it is about people: their feelings, behaviour, dreams and goals. Most of people do not have the analytical habits to check their indexes and charts before taking some action. We realised that they need a reference, a reliable support to advise them about financial decisions.
Aiming to help millions of people, we created Marvin, a virtual assistant backed by machine learning and AI. It motivates you to optimise your consumer behaviour by focusing on your goals and giving you constant feedback and feed forward.
The evolution of Natural Language Processing allows us to build a great user experience, so our users have the ability to talk with Marvin. They can ask questions about their goals, interests, budget and financial projections. Without this technology, we would need to build a much complex UX. Virtual assistants are not a fad, they are not just the future, they are happening right now. But it does not mean that bots will be the only way to interact with your information and services. It is complementary.
Matt Hodgson, CEO, Mosaic Smart Data
Predictive data analytics in capital markets
Predictive analytics is concerned with the prediction of future trends and outcomes using approaches such as machine learning and statistical methodologies. Machine learning techniques have become increasingly popular as a means of generating predictive analytics due to their ability to identify the factors that lead to certain outcomes from within large data sets.
Leading companies in the financial services industry are already using machine learning and predictive analytics to gain new insights into customer behaviour. As a result, they can deepen existing relationships with their customers and build new ones, both leading to increased revenue opportunities. This ranges from the ability to retain clients by identifying those that are at risk of defecting, to providing a more tailored service. Machine learning and predictive analytics are set to become powerful tools for banks’ sales and trading operations.
