AI Powered chatbot using

  • 22 June 2021
  • 1 reply

Userlevel 2

Hi All,

One of my friends is trying to build a AI powered Financial Assistant chatbot.

The intent is for users is to interact with the chat bot via both Voice Commands and Text. User can ask chatbot to calculate their potential monthly car loan payments, mortgage payments, credit card payments, investment returns etc. User can also ask chatbot about current mortgage rates in the market, car loan rates, insurance rates etc.

AI would need to pick data from user’s speech and use it in the calculations. User can also convey the questions and data via texts to AI. AI would then have to perform calculations and talk back to the user to convey the results and also text the results to the user.

If a user asks a question about mortgage payments or car payments, then the AI may have to search the web to provide the best or lowest interest rate to the user and perform the calculations based on that interest rate.

The AI would also look at the user’s spending history and pattern and would suggest how much monthly payments the user can comfortably afford without sacrificing user’s savings goal.

Here are a few scenarios:-

Scenario 1:

User: Can you please calculate the monthly payment if I borrow $400,000.00 for a home loan?

App: Sure. I can help you with that. Can you please tell me how much deposit you can make? Can you make a 20% deposit?

User: Yes, I can make a deposit up to $100,000.00.

App: Great. This means that you can borrow $400,000 with a 20% deposit. This will give you a better rate because your loan to valuation ratio is less than 80% and you are considered a responsible customer by banks.

The best interest rate I can find is 2.25% by If you borrow $400,000.00 for 30 years, then your monthly payments will be $1,528.98. I have also calculated how much interest you can save by paying $50 extra every month. You can save $7,314.56 in interest and pay off your mortgage in 28 years and 7 months.

I have also taken the liberty to calculate your monthly payments and savings based on a 20 year home loan. The monthly payments are $2,071.23. And you will save $53,338.43 in interest.

Based on your monthly income and spending habits, you can afford to take a 20 year home loan to avoid paying extra $53,338.43 in interest.


Scenario 2:

User: Can you tell me about exchange traded funds?

App: Sure, I will be happy to tell you about ETFs. ETFs stand for Exchange traded funds. ETFs are a low- cost way to earn a return similar to an index fund. These are traded in stock market just like shares. These are basically shares of Index Funds. You can buy and sell ETFs through the same way you buy and sell shares. I hope this has been helpful. Do you have any more questions? "


Scenario 3:

User: Can you please help me save more?

App: Sure. Let us take a look at your monthly spending over the last 3 months. Seems like you spend almost 30% of your monthly salary on eating out and entertainment. You can maybe reduce it to 20% so that instead of spending $1,500 on eating out and entertainment, you can spend $1,000. So the $500 you will save, out of this, you can spend $250 on buying additional groceries so you can eat at home. And the remaining $250, you can save. If you want to grow this saving, then you can invest it in a low cost index fund or buy ETFs. In the last 25 years, Australian Stock Market has returned 5 percent on an annual basis. Based on this 5% annual return, if you invest $250 every month over the next 25 years you can increase your savings to 143,181 dollars. Compared to that if you save $250 every month by putting

it under the mattress, then it will increase to only $75,000, which is significantly less than if that amount were to be invested. I hope this has been helpful.


These are a couple of scenarios where the AI chatbot should perform advanced calculations by taking both the text and speech of the user. since we have less time and we have to build a Minimum viable product (MVP) we are hoping to know if powered chatbots can solve these queries.


Best answer by bmunz 25 June 2021, 18:56

View original

This topic has been closed for comments

1 reply

Userlevel 4
Badge +2

Hi Venkat,

The think that’s important to understand with in particular is that our engine specializes in the NLP/NLU side of things.  We don’t have an out-of-box chatbot product, we have NLP capabilities which could be used in building out a larger chatbot product. 

For example, in your first scenario, you can see if you put the question that was asked into our demo/try tool that our disambiguator can help to identify the terms in the question that may be relevant to the bot:

In the larger bot software being built, a developer could use this to best understand what the person is asking for and return the relevant response.  Detecting “calculate,” “home loan,” “monthly payment,” and “$400,000” for example, could be the information needed to program the bot to return the follow up question.

Does that make sense?