Chatbot based robo-advisors — The best AI can do currently to help retail investors? | by sai kiran reddy | Apr, 2021

In my previous post I wrote about robo-advisors. I have explained in detail about the reason for the success of companies offering robo-advisory platforms in the US and discussed about some challenges similar companies face in India. Read the article to understand the current state of robo-advisors better. The link to my previous post can be found here.

Just like stocks now, a few years ago the world of AI made me excited . My current project at Virtusa revolves around optimising the NLP and enhancing the features of chatbots for a consumer banking client. While researching about robo-advisors I learnt that a chatbot based robo-advisor is the intersection of two things I am currently in love with, investing and chatbots.

While studying an Analytics course at University of Hyderabad I built a stock prediction model which had very poor back testing results. After a bit of reading I realised that it is hard to build a good stock prediction model as it is difficult to understand and figure out the causation of things in the market. You can maybe just dump a lot of data and build a deep learning model with amazing back testing results but again it becomes difficult to explain the model and quantify risk. Explainability and risk quantification are very important when you build such financial models. So the best thing AI could do currently to enhance investing for a retail investor is to build an AI based conversational agent that will help them in making their investment decisions. These kind of conversational agents are called chatbot based robo-advisors.

You can look at chatbot based robo-advisors through the eyes of consulting companies that have wealth management clients or mutual fund companies that are planning to offer investment advisory services.

Consulting firms :

Since I work in a B2B consulting company, I will first give my perspective on chatbot based robo-advisory solutions with respect to such companies. Any AI/ML solution built by a B2B provider is going to either reduce costs or increase revenue for any client. Chatbot based robo-advisors fall under this unique category of achieving both the objectives for your clients in the area of wealth management.

1. Firstly, they save costs because clients don’t have to hire multiple human fund advisors to manage their customers, a single chatbot can mange all the customers.
2. Secondly, if you build a very good chatbot with a nice user experience and algorithms within the bot are good enough to recommend the right funds to the customers, current set of customers of your client are more likely to recommend the platform to other new people and in turn also help increase the revenue for the client.

1. Chatbot Trends Report 2021

2. DO’s and 3 DON’Ts for Training a Chatbot NLP Model

3. Concierge Bot: Handle Multiple Chatbots from One Chat Screen

4. An expert system: Conversational AI Vs Chatbots

Therefore, it makes sense for a B2B provider looking to increase their foot print in the AI space to focus on helping their wealth management clients build good chatbots that give investment advise.

I feel many wealth management companies in the future are going to have their own version of a chatbot as a robo-advisor because chatbots offer more interactivity with the customer and can make the overall user experience way better. Some big banks already have such chatbots but they are not sophisticated enough. Good machine learning based chatbots are still in a nascent stage and so much more can be done to make the interaction with chatbots more human like.

Mutual Fund Investment Platforms :

I believe robo-advisors in the form of a chatbot is the right way even for companies that let you invest in mutual funds in India to offer robo-advisory solutions. As I mentioned in my previous post many mutual fund platforms in India are offering many services for free. One obvious way such companies would generate cash in the future is by providing investment advisory solutions. The cheapest way for them to provide personalised investment advise to their customers with a good user experience is through such chatbots that act as robo-advisors. The data that such companies collect about the customers can be used to train the chatbot to have more personalised and interesting conversations with customers. This could turn out to be a big competitive advantage for mutual fund investment platforms.

Features we can build using chatbot platforms currently available in the market –

I would like to mention some good to have features of a chatbot that helps the customers, of a wealth management client of B2B consulting companies and any fin-tech start up or mutual fund investment platform that wants to offer investment advisory services, in making investment decisions.

• The chatbot should gauge the risk appetite of customers using the platform and understand their financial goals.
• Do asset allocation and construct a portfolio based on the customers risk appetite and financial goals.
• Rebalance the portfolio when the financial goals of the customer change or the macro economic conditions change.
• Answer questions around the customers retirement planning and other financial goals like child education etc. For example it should give answers to question like ‘how much should I save every month to accumulate x amount during my retirement’.
• One feature I would love to see is that the chatbot should educate the customer and allow him to modify his portfolio based on the knowledge he has gained. The thrill of knowing why you are investing in something is much more than a bot just recommending something without letting you know the ‘why’. Even if the customer doesn’t want to modify his portfolio as an investor he would feel safer and more comfortable to use the platform if he knows why a bot is investing in a particular fund.

Most of the current set of AI based chatbots have an intent classifier that tries to identify the intent of the customer and an NER ( Named entity recognition) model to identify entities that are required to fulfil a particular task. The dialogue management is mostly rule based. All the good to have features mentioned above can be implemented using such chatbots. Looking at the pace at which NLP developments are happening, I have a feeling that we would find better ways to use AI to build more powerful chatbots very soon and the features chatbot based robo-advisors will have are only going to get better.

from
https://fatsfixedassettracking.com/chatbot-based-robo-advisors-the-best-ai-can-do-currently-to-help-retail-investors-by-sai-kiran-reddy-apr-2021/2020/