Technology is paving the way for organisations to stay relevant in the ever-evolving business landscape. We are streamlining our core business functions such as sales and logistics using digital technologies, looking at the strategic objectives of each business.
We began our digital transformation journey about seven years ago, and today have sharpened our focus on the use of digital technologies such as artificial intelligence (AI), machine learning (ML), cloud computing, internet of things (IoT), automation, and advanced data analytics to empower decision-making, and improve business processes and customer experiences.
We are leveraging our High Tech + High Touch strategy to serve budget customers in Tier 2&3 cities of India.
As we expand our presence in Tier 2&3 cities of “Bharat”, the Innovation Lab is a significant element of our growth strategy aimed at accelerating digital lending transformation for “Bharat”.
Piramal Innovation Lab’, a state-of-the-art Centre of Excellence for Technology and Business Intelligence in Bengaluru is helping us create a vibrant ecosystem of fintechs, start-ups and tech innovators. Our initiative of setting up this lab is aimed at accelerating the development of next-generation lending solutions and analytics for the unserved and under-served “Bharat”, and aligning with how our consumers are reimagining the industry.
We are investing significantly to become a data-led and AI-first organisation, which reflects in every aspect of our way of doing business. We operate at the intersection of finance and technology, leveraging advanced artificial intelligence algorithms and machine learning models as the cornerstone for its lending practices.
We are increasingly focussing on how we can make our business decisions based on data analytics. We make use of AI, ML, decision science, and automated business intelligence to improve customer experiences and detect fraud and delinquency. We have established newer ways of working including agile methodology, self-empowered scrum teams, design thinking, collective problem-solving, leverage idea generation, and empowering teams to contribute to the organisation’s growth and success.
Our focus is shifting from automation and cost efficiency to more advanced analytics and using AI/ML for business decision-making. A lot of financial processes that we earlier started with RPA are now maturing to AI/ML to make it more automated and human-less. We are leveraging AI/ML across the customer space for risk management and fraud prevention and to improve employee experience and reduce attrition.
Following introduction of Information Security framework for NBFCs in June-2017 by RBI/NHB’s, we revamped our Information & Cybersecurity journey. Recognising the evolving threat landscape and ever-increasing importance of cybersecurity, maintaining the highest standards of information security became the topmost priority for Piramal Enterprises.
Initially, we leveraged Gartner Cybersecurity Maturity Framework that served as the foundation for our subsequent actions. As part of this, we periodically engaged external experts and consultants to conduct an ethical hacking exercise (Red Team assessment) to identify vulnerabilities and areas for improvement across our systems, processes, and infrastructure. The insights gained from these assessments helped us refine our security controls and ensure that we are proactively addressing potential risks.
To address the findings, we focussed on technology deployment to fortify our infosec infrastructure. This included implementing advanced security solutions and leveraging cutting-edge technologies like Advance Malware Solution (EDR), Web Application Firewall, Privilege Access Management, NextGen Security Monitoring & Threat Hunting and Domain Reputation.
In FY 2023-24, we conducted the Cybersecurity Maturity Assessment against Global NIST Cybersecurity Framework, followed by leading global banks, and defined a three-year comprehensive cybersecurity roadmap. We identified areas of strength and areas that require improvement by aligning their practices with the framework’s guidelines, enabling us to implement appropriate cybersecurity measures in line with leading banks.
We also recognise that our employees are a critical line of defence against cyber threats. For strengthening our organisation’s resilience to cyber-attacks and equip our employees with the necessary knowledge and skills to mitigate security risks and adhere to our established policies and procedures, we continue to implement comprehensive cybersecurity awareness campaigns in form of gamified cybersecurity awareness programmes, cyber quiz, posters and email communications, periodic training sessions on various facets of cybersecurity. To assess the awareness of our employees, we initiated the process of conducting periodic phishing assessments.
We have instilled the practice of democratising insights and promoting data driven decisions throughout customer life cycle. Some of these key highlights are follows:
AI-Driven Credit Decision Models
The core risk models reflect the core tenet of our organisation i.e., data-driven decision-making. Despite the presence of standard risk scorecards in the market, we built custom scorecards for each of our product offerings. The market scorecards follow a one-size-fits-all approach & for this very reason, they are built in such a way that they work for all secured and unsecured products alike, thus diluting the scorecard performance. On the other hand, product-specific scorecards allow for identifying behaviours that are unique to each product and help in predicting the chances of default. Such behaviours will be overlooked by the market scorecard as they may not be applicable across all products.
As part of our core risk models, an applicant’s credit report is comprehensively analysed with the assistance of 1,500+ variables derived from them. An industry standard sophisticated pipeline was designed for identifying top risk predictors. Millions of industry data points were used to train models using state-of-the-art algorithms to arrive at a model stable across different periods. Most of these custom scorecards have had a 1.4x performance uplift when compared to the market scorecards, thereby signifying the essence of having custom scorecards built.
We have developed tailored scorecards for each of our secured and unsecured products. These scorecards are designed to account for various customer behaviours within each product category. For instance, we recognise that customers who choose digital over physical personal loans exhibit distinct characteristics that warrant specific considerations. Similarly, customers seeking larger loans, such as ₹ 10 lakh compared to ₹ 50k loans, display different behaviours throughout the loan lifecycle. Consequently, we customise our acquisition approaches to accommodate these differences, ensuring that each customer receives appropriate treatment aligned with their unique needs and behaviours.
We have also worked on an innovative approach that seeks to bring objectivity to policy decisioning. Typically, several policy univariates are applied in the BRE (Business Rule Engine) to arrive at a recommendation for the underwriter. However, the univariate nature of these rules cut out a sizeable chunk of the approved applications thereby bringing down the approval rates. The univariate nature applied across varied, diverse documents and information available for a customer often makes the entire process subjective to the underwriter handling the case and takes a lot of time for them to analyse everything available and make the decision accordingly.
Therefore, we developed multivariate AI models that consider a combination of multiple variables in arriving at the recommendations. This allowed us to swap in a lower risk base, thereby ensuring higher approval rates. At the same time, it also allowed us to make STP decisions as well as assist underwriters in making sense of multiple attributes before making the final decision of sanction/reject. This has helped us reduce TAT and has helped us bring uniformity in decisions.
Understanding a customer’s capacity to pay is a critical step in underwriting. As an NBFC, where we do not have direct access to a customer’s banking data, we must rely on other sources of information to determine an applicant’s capacity. There are various sources of such information, be it bank statements, salary slips, income tax returns (Form 26AS), GST filings, EPFO deductions, etc.
As the initial step into building our digital income assessment capabilities, we have taken up the detection of salaries in bank statements. We deployed a diverse set of machine learning algorithms such as clustering methods, natural language processing, and unsupervised learning to improve salary detection in bank statements.
We are now capable of increasing digital income detection, through our internal bank statement analyser, enabling customers to undergo digital assessment without the requirements of a physical underwriter. This frees up the bandwidth of underwriters to look at more complex cases, provides operating leverage and improves customer experience as the time to disbursal is reduced. The product is live in decisioning for our Digital Personal loans and has improved income detection by 2x.
While we have seen improvements in salary detection, this is only a part of the capabilities that are built. The information that is extracted from the bank statements also enables us to check for fraud patterns in the transactions, irregularity of salary as well as other information that would be useful for underwriting, such as minimum balance maintained, presence of other loans and bounces in those transactions.
The recent increase in coverage of digitised banking data through account aggregators enables customers to have a seamless process when applying for loans, without the hassle of uploading multiple documents, while improving the authenticity of the data. This extracted information will complement the bureau data used in the credit risk models, which will help improve approval rates while managing risk.
Piramal Finance has taken a bold leap forward by introducing a groundbreaking GenAI Chatbot, exclusively tailored for its sales teams. This state-of-the-art tool has sparked a paradigm shift in productivity enhancement strategies, seamlessly integrating into the daily workflows of sales professionals and revolutionising their approach to accessing critical sales data.
Leveraging the advanced capabilities of Language Model Machines (LLMs), the GenAI Chatbot offers a conversational interface to sales professionals, granting instant access to crucial performance metrics. By eliminating the need for manual calculations and data retrieval, this innovative solution empowers team members to redirect their focus towards strategic decision-making and revenue-generating activities.
The Chatbot’s robust security measures always ensure the protection of sensitive sales data, allowing authorised users to securely access it from any device, be it a mobile phone, tablet, or laptop. This unparalleled flexibility and convenience come without compromising on data security, reinforcing Piramal Finance’s commitment to safeguarding confidential information.
Furthermore, the Chatbot’s capability to present data in visually appealing and easily understandable formats enhances user engagement and comprehension. Sales professionals can swiftly extract actionable insights, enabling them to devise more effective sales strategies and capitalise on emerging opportunities.
Through its intuitive interface and seamless data retrieval capabilities, the GenAI Chatbot empowers sales teams to operate with unprecedented efficiency and precision. By harnessing data-driven insights, individuals can optimise their performance, monitor progress towards targets, and drive higher levels of sales productivity across the organisation.
Piramal Enterprises’ introduction of the GenAI Chatbot represents a significant milestone in the evolution of sales productivity enhancement within the financial sector. It not only underscores the Company’s commitment to innovation but also sets a new standard for leveraging AI-driven solutions to achieve tangible business outcomes.
A human touch is indispensable even in a technology-driven world. With a slew of exciting and innovative measures that leverage the powerful and growing capabilities of advanced Generative AI, we aim to streamline and optimise our customer-facing activities. Let us take a quick look at current initiatives and what is in store.
The CX team currently has a fully functional interactive dashboard to understand the voice of the customer, named Dhwani. This tool primarily tracks key issues and trends customers encounter at a broad level, offering senior leadership and middle management immediate insights. The data is based on actual customer voice calls, emails, and branch visits, making the view as holistic as it can be. This data is analysed using advanced Machine Learning algorithms, while leveraging the power of Generative AI to provide succinct summaries of the key issues currently being faced by customers along with the emerging trends.
Our Contact Centre gets thousands of calls every day and our agents help our customers address their problems. But from a company’s perspective, we would want to have a good call governance process to ensure higher customer satisfaction. This process was handled manually, reducing the coverage while involving a lot of subjectivity. This process was automated using advanced NLP while using the power of Generative AI to govern the aspects of call quality monitoring and providing relevant feedback to individual agents. The feedback is then passed on to each agent to help improve their customer addressal process.
Post Call Analytics dashboard view:
In conjunction with noted AI lab ThoughtWorks, we have developed an interactive voice-bot, which builds on state-of-the-art real-time audio transcription and translation pipelines and Large Language Models (LLMs), to assist prospective customers in completing their loan journey. The bot can understand and provide relevant replies to a variety of customer queries, with minimal latency. This asset is currently deployed as a POC and a roll-out is planned soon for a few product journeys.
For an organisation in the business of lending with 7,000+ relationship managers and hundreds of credit underwriters, it is imperative to ensure data availability at the grassroot level. Hence, at Piramal, we worked on the vision of democratising data. Data democratisation provides easy and secure access to real-time insights, fostering data-driven decisions and collaboration. Teams leverage objective information for discussions which eventually lead to better choices, ownership, and organisational alignment.
To achieve these goals, Piramal has meticulously captured key KPIs across business initiatives. These KPIs are presented on dashboards and reports tailored for audiences ranging from senior leadership teams (SLTs) to frontline team members like relationship managers (RMs). Our daily sales huddles, credit underwriter performance management calls, and monthly function/business reviews (MFRs/MBRs) with the CEO all leverage our dashboards as the only source of truth.
Every function and product has a dedicated dashboard with near real-time access to performance metrics. This enables teams to monitor performance and adjust strategies based on the ground situation in near real-time. Compared to competitors in the BFSI industry who rely on T-1 data reporting, our near real-time dashboards provide a significant competitive advantage. These solutions power our proactive strategies, enabling Piramal to seize opportunities, mitigate risks, and solidify our industry leadership by making data-driven decisions in a dynamic market.
Copyright 2024 @ Piramal | All Rights Reserved.
