A password will be e-mailed to you.

In Interview with Pravin Lal, CEO, and Founder Capital Quant Solutions

Pravin Lal, CEO, and Founder Capital Quant Solutions

1. How did you come across the idea for Finstinct?

While I was working with Lehman Brothers, we had a portal called Lehman Live which had a lot of research reports and other content available as PDF documents for the clients on the street. These reports were used by the clients for taking important investment decisions. It was a very content rich application and at times I used to read through some of the reports on Lehman Live. At that time, I used to wonder how would people consume so much of information that is there on this portal. It was not humanly possible to do so but there was wealth of information on Lehman Live. 

Around the same time i.e. before 2008 there was a company called Gridstone. They were solving the problem of information overload from documents for the financial services industry and they had a team of people who would extract the details and structure them and make it available for analysis. Gridstone was obviously adding value as it had clients on the wall street who were subscribing to their service. However, this was done largely manually by the Gridstone team as that was not the time when machine learning, NLP and AI had found commercial usage. 

Around 2015, we started witnessing commercial usage of ML and AI in financial services. That was the time when I decided to start a cognitive automation product company which would solve the problem.

All these thoughts led us at Capital Quant Solutions to build FinStinct which is a cognitive automation product focused on the financial services industry.

2. The role that Finstinct is playing in financial institutions like NSE, SmartStream, etc. and how the B2B automation space is shaping up?

Most functions in a financial services organization are document intensive especially when you look at organizations like a stock exchange where there are documents coming from the companies listed on the exchange as well as the member brokers. These documents could be in the form of financial results of companies, corporate announcements, DRHP/RHPs, legal undertakings, bank guarantees, member certifications etc. Processing the details in these documents includes extracting the relevant details from the document and in some cases performing an automated checklist running on these documents or even performing reconciliation of information provided in the documents. These processes are executed automatically by FinStinct.

At NSE, we are working with various divisions within the exchange to extract key value pairs from documents and also perform reconciliation of information in these documents. We are working with Smartstream to pull out information from multiple global stock exchanges and we provide them with an automatically generated content repository created by extracting key value pairs. 

The B2B automation space is at an inflection point now.  There is a growing demand to move towards intelligent or cognitive automation to reduce cost and enhance efficiency. The ROI from such initiatives is offering a very compelling reason for organizations to be serious about the intelligent automation initiatives. In addition, the pandemic has also resulted in a higher momentum for adoption of such initiatives. Pre pandemic there was resistance in some form to accept the change led by AI/ML. However, the past one and a half years have made it clear that over the coming decade the financial services industry will experience an AI/ML led disruption in processes which deal with unstructured data. 

As per the WEF report by 2025, the time spent on current tasks at work by humans and machines will be equal. The report further estimates that by 2025, 85 million jobs may be displaced by a shift in the division of labor between humans and machines, while 97 million new roles may emerge that are more adapted to the new division of labor between humans, machines and algorithms. 90% of the financial services companies surveyed by the WEF future of jobs survey indicated that AI technologies are likely to be adopted by 2025. 

3. Who are your market competitors and target audience?

We are a B2B company focused on the financial services industry. We cater to banks, NBFCs, brokerage houses, asset management companies, rating agencies, insurance companies and stock exchanges. 

As we see the competitive landscape in this space can be viewed in three parts. First, there are the global tech giants who offer generic solutions which are domain agnostic. Their offerings are available on the cloud. However, the limitation is that since their offering is domain agnostic, they do not offer models which are specific to the financial services industry as a result offer lesser accuracy compared to models which are trained for specific document types. Also, it is important to note that their offering is available only on the cloud and they are not in the on-premise business. Many financial institutions are concerned about the security of their documents which could be sensitive and confidential documents such as financial contracts or legal agreements etc. 

Secondly, there are startups in this space like ourselves which offer a niche product which is focused on the domain. There are domain centric models which come with the solution and we offer both on premise and on cloud. We see a lot of interest within the target segment to move ahead with such niche organizations. Here it is important to note that most startups are offering a point solution which means that they have a solution for specific document type such as invoices and bank account statements. There are very few niche organizations which offer a comprehensive cognitive automation solution focused on the financial services industry. FinStinct can work with any document as it offers the FinStinct product which comes with pre-configured models for specific document types and the FinStinct platform which offers a very powerful DIY engine which empowers a user who does not know machine learning to build a machine learning model to extract specific data from any document type. 

Lastly, we have the KPOs and captive setups of banks which are an indirect competition to us as they offer the same service to their clients but they hire people who do the work manually. Over the last decade these organizations grew exponentially with the outsourcing of work from financial services organizations. However, we believe that this space is up for disruption. Either some of the KPOs will acquire the second segment of players or it will be the other way round.   

4. How do your products make work easier?

Financial services industry relies heavily on information available in complex and unstructured documents for decision making as well as for processing various transactions and operations. The work entails extraction of information from documents, structuring the same and then analysing of the extracted information. The extraction could be in the form of key value pairs or even the extraction of tabular/textual data. 

This is a requirement in all divisions/functions within our client segments from equity, currency, commodity to credit, trade finance, legal and compliance etc. Currently this is accomplished by manually reading through the documents and identifying the relevant information which is then extracted into an excel and then analyzed. The manual process has the following problems:

  • It requires expensive human bandwidth and is time consuming. To understand the jargon in the documents you need people who are familiar with the processes in a financial services firm. Typically, it is an MBA in finance or a CA who is reading through these documents. For eg generating credit spreads from the financial results of a company is a function performed by a credit analyst. Typically, a junior credit analyst would take around 2 days to do the spreads of a company while a seasoned credit analyst would do the same job within 2 to 3 hours. With FinStinct the same can be done within 10 min.
  • Manual process is error prone as errors of omission can always creep inside.
  • The human mind has a limitation of scale. It can only read through a limited amount of information from unstructured sources. Coupled with the fact that the amount of information from such sources is increasing and it is increasing exponentially that poses another challenge.

FinStinct addresses the above problems by offering a comprehensive solution which can work on any document type. It brings in the process efficiency and offers an average ROI of 40% to our client organizations.

5. What are the specific challenges your tools help address with unstructured data?

We offer a solution which not only reduces the cost of operations and enhances the efficiency/productivity but it also helps in reducing the risk of taking sub optimal decisions as the human mind has the limitation of scale and is not able to process the voluminous information coming from the unstructured sources. While it is easy to estimate the cost saving of a process, the cost of reducing the sub optimal decisions is anybody’s guess. It is estimated that an average knowledge worker takes 2.5 hours per day just searching for information from unstructured sources. This is greatly reduced by using FinStinct. The cost saving in this case is an easy math. However, sub optimal decisions that gets taken at the time of underwriting can be expensive and hard to estimate. The cost saving of such sub optimal decision is much more than the cost saving by making the process more efficient. FinStinct addresses both sides of the challenge. 

6. Are there any new products in the line?

Absolutely, we are coming with at least two more products which we believe would be disruptive in their space. Both would rely on the unstructured data processing for the financial services industry. We would be leveraging the work done by FinStinct to launch these two products. 

7. What is your vision for Capital Quant Solutions?

We want to be a global AI product from India for the financial services industry. All our offerings will leverage AI, ML and NLP and will be productized offerings for the financial services industry globally. We believe that the status quo in the financial services industry globally is being challenged by adoption of cognitive technologies and we want to be at the center of this disruption.

Comments

comments

No more articles
Send this to a friend