How intelligent automation helps the decision-making process

Vaibhav Gupta of Visionet Systems explains how implementing intelligent automation solutions can improve efficiency within a financial institution and improve its end-customer’s experience as well

Vaibhav Gupta, Head of Products, Visionet Systems
Vaibhav Gupta, Head of Products, Visionet Systems

A lot of banking processes today require images of documents to be managed. These processes include mortgage applications, KYC, insurance, etc. Intelligent automation brings the capability to identify a document, read and extract the required data from that document and prepare the data ready for decision-making. According to Vaibhav Gupta, Head of Products at Visionet Systems, this leads to better decision-making of a proposal with faster processing times and fewer errors.

Gupta has worked extensively in leading the product journeys for some of the world’s top lending solutions. At Visionet, he heads the product and business strategy for the entire banking and financial services portfolio with the intent to bring an innovation-driven approach and create more impact in the industry. He explained how intelligent automation goes beyond robotic process automation (RPA):

“RPA is rules-based automation of low-level tasks to increase speed and reduce errors. Intelligent automation uses a lot of different technologies, such as computer vision, OCR and AI. Intelligent automation can hence take the next step in automation by bringing capabilities that are more complex than RPA. An example is reading documents and extracting the required information.”

How do you identify opportunities to monetise Intelligent analytics, and how does this affect business strategy and product development?

“Functions in banking that have large human resources are a quick way to shortlist use cases for AI and ML. Today a lot of work that we do can be assisted by AI and ML. This increases the quality, improves the work culture of the team, and enables you to take up initiatives that can increase the value that your teams bring to the organisation.

“Banks and financial institutions can now move closer to ‘no touch’ banking facilities even to complicated and high-profit transactions such as mortgage and Insurance claims. This increases scalability and brings a lot of customer satisfaction at the same time. Delighting your customer and having the ability to fulfil an increase in demand while reducing your cost of operations is a really good way to get ahead!”

What are the 3 key areas in intelligent analytics in banking and finance?

“I would try and answer this in a different way. Look at the functions with the following lenses:

  • Descriptive Analytics: What does the data say? – e.g., insurance claims processing
  • Diagnostic Analytics: What went wrong? – e.g., identifying default behaviour in loan segments
  • Predictive Analytics: What do you think will happen? – e.g., mortgage underwriting
  • Prescriptive Analytics: Which automated actions must happen if something goes wrong?

“Document processing is an area that should be looked at as the next step in automation using analytics. The improvements in customer satisfaction and cost savings are significant. OCR, computer vision and AI work together to bring a very robust solution for document processing. DocVu.AI’s mortgage solution does the straight-through-processing of 27% of document types in a mortgage application. Even if the document format is non-standard, DocVu.AI can process this with its template-less feature. The technology is mature to implement.”

 Customers experience the benefits of AI… through, for example, faster processing of mortgage loan applications, instant insurance claims and more accurate loan underwriting

How complex/costly is it to create and manage new Intelligent analytics solutions?

“While this is a very complex question, a quick way to evaluate is: are there solutions available that can meet your functional and organisational requirements, or do you need to build a custom solution to address this? If the answer is the former, then the complexity and cost are fairly low, and implementation times are around 30 days. However, if the answer is the latter, then the implementation could take up to 6 months.

“Let’s take document processing as an example. Building a boutique solution could take a year to operationalise but taking a solution that already addresses this would take less than a month and, if the deal is structured correctly, would also not have any implementation costs. A couple of things that you should consider are: is the solution a generic solution or is it designed for that specific function, and what experience does the vendor/partner have in your specific industry?”

What are the basic building blocks of successful data management?

“As data is critical for intelligent analytics, the objective of data management has evolved. In the past, data management revolved around access to data and storage of data but today, data management brings a whole lot more to the table. Data management provides inferences and insights about customers, operations, sales, products, etc. So, the basic building blocks revolve around how you get these inferences.

“The first 2 steps are called the build phase:

  • The first step is ingestion techniques. How do you ingest the right data for your objectives?
  • Identify, build, and test the data models required to get the inferences and insights from your data.

“The next steps are the operations phase:

  • Data access and curation – ensure that the availability of the data required is steady and does not change in quality.
  • Quality Checks – check all the inputs and outputs of the data models. This is paramount as your data might go beyond the limits of the data models designed for you and could lead to bad inferences.

“One key thing that we always need to consider is that we should know our objectives and our data to build a genuinely rewarding data management system.”

How do you maximise the value of intelligent analytics through a curated approach?

“Accurately define your objective, understand the causal relationship between data patterns and the objective and select the right models. While this can get very technical at times, the critical part is very logical and getting that right is very important.

“Once this is established, validate, validate and validate during design and don’t forget to validate when the system is in operation as well!”

How do you define critical success factors and, equally as important, identify and resolve the potential failure points?

“Identify bottlenecks in your processes and see how the intelligent automation solution resolves these. The bottlenecks could be resource constraints or data complexity. If you have a plan to effectively address these 2 aspects in your process, then you are heading towards success.

“Look for precision from the intelligent automation solution, which can only come from data validation and ensure the quality of data processed by the intelligent automation solution. Sometimes, to improve the quality of data provided to the intelligent automation solution, some changes could be required in the process.

“One needs to be careful that objectives and relationships are crystal clear and that the data is available to realise this. The next aspect to consider is the right analytics. These are issues that can lead of lengthy delays in projects.”

What does intelligent automation mean for end customers’ experience?

“The key areas that Intelligent Analytics can benefit customer experience are:

  • Reduced touch points
  • More accurate assessment
  • Real-time execution of complicated processes

“Customers experience the benefits of AI directly or indirectly in financial institutions through, for example, faster processing of mortgage loan applications, instant insurance claims and more accurate loan underwriting (which leads to better loan proposals). It doesn’t matter whether the implementation is in operations, customer acquisition or support.

“In mortgage loans and insurance claims, document processing takes up significant time and effort. Streamlining this so that it makes the process faster and reduces mid-process follow-ups is critical in improving an end customer’s experience. That’s what DocVu.AI’s document processing brings to mortgage, insurance, and finance & accounting – accurate financial document classification and processing!”