In an exclusive interaction with Express Computer, Dr. Debarag Banerjee, Chief AI & Data Officer, L&T Finance discusses how L&T Finance is building an AI-native lending organisation, the principles behind its flagship AI platforms, and why the future of financial services will increasingly be driven by autonomous, yet responsibly governed, AI systems. Dr. Banerjee also highlights how the company has embedded AI across the lending lifecycle, from underwriting and portfolio monitoring to customer engagement and collections, while delivering measurable improvements in credit quality, operational efficiency, and financial inclusion.
Artificial Intelligence (AI) is rapidly becoming the defining competitive advantage in financial services, but only a handful of organisations have successfully moved beyond isolated pilots to enterprise-wide transformation. L&T Finance is among those leading the shift. In fact, AI has moved well beyond experimentation at L&T Finance. Rather than viewing AI as another layer of automation, the company has positioned it as the foundation of its operating model, driving better lending decisions, improving customer experience, reducing operational costs, and creating new avenues for business growth.
For Dr. Debarag Banerjee, Chief AI & Data Officer at L&T Finance, AI today represents much more than operational efficiency.
“AI is simultaneously lowering credit costs, reducing collection expenses, automating operations, and creating entirely new AI-native growth opportunities. Together, these capabilities become a strategic differentiator that enables profitable, risk-adjusted growth,” says Dr. Banerjee.
This vision has steadily evolved since the company began reimagining the future of lending through AI in late 2023. As large language models matured, Agentic AI emerged, and regulatory understanding deepened, L&T Finance continuously refined its strategy to ensure innovation remained aligned with business outcomes and governance.
Scaling AI beyond PoC
While many organisations continue to struggle with taking AI initiatives beyond pilot projects, L&T Finance has deliberately shifted its philosophy. Instead of pursuing proof-of-concept projects simply to showcase technological capability, the company now evaluates every AI initiative through the lens of production readiness and enterprise adoption.
According to Dr. Banerjee, successful AI deployment in financial services requires far more than developing accurate machine learning models. It demands deep collaboration across data engineering, IT, application teams, cyber security, credit, risk, legal, compliance, and business functions.
This cross-functional execution framework has enabled L&T Finance to design, deploy and scale enterprise AI solutions within two quarters, a pace that few regulated financial institutions have been able to achieve consistently.
Building an AI ecosystem across the lending lifecycle
Rather than developing isolated AI applications, L&T Finance has built an interconnected ecosystem of AI platforms that collectively span the entire customer journey.
Project Cyclops serves as the company’s digital underwriting intelligence engine, automating credit decisions in near real time. Project Nostradamus continuously monitors customer portfolios after loan disbursement by analysing hundreds of behavioural variables every month. Project Helios functions as an AI co-pilot for underwriters, improving productivity and accelerating decision-making, while Project Orion provides portfolio intelligence through Agentic AI. KAI brings conversational intelligence to customer interactions through multilingual chat and voice AI capabilities.
Together, these initiatives share several common architectural principles.
The company combines traditional credit data with alternative information sources, including banking behaviour, GST records, geospatial signals, trust indicators and satellite imagery, to build a richer understanding of customer creditworthiness. At the same time, every data source is evaluated using what L&T Finance calls Return on Data Investment (RODI), ensuring the business value generated exceeds the cost of acquiring the data.
“Speed is another defining characteristic. AI-powered underwriting decisions that once required hours can now be completed within seconds, while SME underwriting turnaround times have been reduced significantly,” adds Dr. Banerjee.
Equally important is the company’s emphasis on human-AI collaboration rather than complete replacement. AI systems automate repetitive work, allowing human experts to focus on complex risk assessment, exception handling and continuous model improvement.
Demonstrating measurable business value
One of the biggest challenges enterprises face is demonstrating tangible returns on AI investments. L&T Finance has attempted to address this by publicly tracking business metrics that directly reflect AI outcomes.
Its AI-powered underwriting engine has significantly improved portfolio quality in two-wheeler and farm finance while substantially reducing early delinquencies. AI-driven voice bots are improving repayment behaviour among borrowers, while underwriting productivity has improved through faster loan processing.
The organisation is also witnessing broader enterprise adoption of AI, with growing utilisation of large language models, increasing volumes of AI-ready enterprise data, and more than 100 AI initiatives progressing through various stages of deployment.
Dr. Banerjee further states, “We measure AI success through enterprise value, which includes cost savings, faster decisions, better underwriting, improved adoption, and ultimately stronger business outcomes.”
Towards autonomous lending
L&T Finance has already crossed an important milestone by enabling fully autonomous credit decisions for several high-volume retail lending products.
In categories such as two-wheeler finance, farm equipment finance and personal loans, Project Cyclops independently verifies customer information, evaluates risk and delivers binding lending decisions within seconds.
However, autonomy is accompanied by strong governance. Applications that fall outside predefined confidence thresholds are automatically escalated to human experts, ensuring that AI operates within clearly defined risk boundaries while allowing experienced professionals to focus on exceptional cases rather than repetitive processing.
Financial inclusion through AI
Perhaps the most significant impact of AI at L&T Finance lies in expanding financial inclusion.
Traditional credit bureau data often fails to adequately assess India’s large population of first-time borrowers, self-employed professionals and rural customers. By incorporating alternative data sources and behavioural indicators, the company has been able to build robust underwriting models for customers with limited formal credit histories.
This approach has enabled significant improvements in underwriting accuracy for new-to-credit customers while simultaneously supporting business growth without compromising portfolio quality.
“Our most meaningful AI achievement has been our ability to safely and profitably underwrite India’s new-to-credit segments using alternative data as reliable proxies for income and repayment capacity,” points out Dr. Banerjee.
Data – the foundation of enterprise AI
Despite growing excitement around generative AI, Dr. Banerjee believes successful AI strategies begin with data rather than algorithms.
L&T Finance has invested heavily in a cloud-native enterprise data platform capable of supporting more than 280 million customer records through real-time data pipelines, robust governance frameworks and consent-based data management.
The company has also strengthened data quality controls, established enterprise-wide stewardship programs, and embedded privacy and access governance directly into the data layer, thus creating a strong foundation for both AI scalability and responsible AI deployment.
As Dr. Banerjee aptly puts it, “While amateurs talk AI, professionals think data.”
Responsible AI as a competitive advantage
Operating within one of the country’s most regulated industries requires innovation to coexist with accountability.
L&T Finance maintains strict separation between experimental AI environments and production systems, while ensuring every production model remains transparent, auditable and explainable.
Consent-based data usage, bias monitoring, regulatory compliance and explainable decision-making are embedded into every AI deployment, enabling innovation without compromising customer trust or regulatory expectations.
What lies ahead?
Looking ahead, Dr. Banerjee believes Agentic AI will fundamentally reshape financial services over the next three to five years. AI agents will increasingly negotiate, orchestrate and execute financial transactions, while advances in Edge AI and quantum computing will unlock entirely new possibilities for privacy-preserving AI and large-scale portfolio optimization.
At the same time, he believes regulatory frameworks will also need to evolve, from focusing solely on risk mitigation to embracing AI-first principles that enable innovation, while protecting consumers and ensuring systemic stability.
For L&T Finance, the objective is no longer simply adopting AI. It is building an AI-native financial institution where intelligence is embedded across every process, every decision and every customer interaction, transforming AI from a technology initiative into a sustainable competitive advantage.