For AI to work in housing finance, it has to be accurate, explainable and empathetic: Kalyan Josyula, Co-Founder & CTO, BASIC Home Loan
India’s housing finance ecosystem is undergoing a profound transformation, driven by artificial intelligence, cloud-native platforms and data-led decision-making. BASIC Home Loan sits at the centre of this shift, having built its digital foundation well before AI entered the mainstream of lending conversations.
In an exclusive interaction with Express Computer, Kalyan Josyula, co-founder and CTO of BASIC Home Loan, outlines how their early technology choices, like microservices architecture, cloud-native scalability and security-by-design, created the backbone for HOM-i, India’s first AI-powered home loan assistant platform. Designed to address long-standing inefficiencies such as fragmented processes, manual verification and low transparency, HOM-i introduces intelligence across eligibility assessment, document verification, lender matching and borrower engagement.
BASIC Home Loan has been focused on digitising India’s home loan ecosystem. From a CTO’s perspective, what were the foundational technology principles that guided the design, scalability, and reliability of your mortgage platform before HOM-i came into play?
Our goal was to simplify one of the most complex financial journeys for Indian consumers – buying a home. We build our platform on a microservices-based architecture to ensure flexibility and seamless integration. Cloud-native setup allowed us to scale rapidly while maintaining pace and performance.
Reliability and trust are very important. We embedded security and compliance at every level, from data encryption and consent-based APIs to continuous monitoring frameworks; maintaining consumer trust is non-negotiable for us.
Our core principles not only powered our initial success but also created a strong foundation for HOM-i, enabling us to introduce AI-driven automation and smarter decision-making on top of a stable, future-ready infrastructure.
According to you, HOM-i is India’s first AI-powered home loan assistant platform. What inspired its creation, and what specific gaps or inefficiencies in the home loan journey were you aiming to solve through AI-driven automation?
The inspiration behind HOM-i came from a very real problem we observed across the ecosystem. While home loans in India had gone digital on the surface, the real processes were still mostly manual, slow and fragmented. Consumers were stuck with approval delays, cumbersome documentation processes and limited visibility into their loan status.
We wanted to change that by bringing real intelligence and transparency into the home loan journey. We wanted to make applying for a home loan as simple and intuitive as applying for a credit card, powered by data.
HOM-i is built to transform a traditionally slow, opaque process into a fast, intelligent and customer-first experience, redefining how India accesses home finance. The AI intelligence behind HOM-i automates key parts of the process, from eligibility scoring and document verification to real-time lender matching. This can significantly reduce manual intervention and turnaround time. HOM-i is continuously learning from user behaviour and lender outcomes, making its recommendations more precise, smarter and personalised over time.
Could you elaborate on the AI and generative AI architecture powering HOM-i? How do the machine learning models handle multilingual voice, text, and video interactions across 30+ Indian languages and ensure accuracy, inclusivity, and data security?
HOM-i’s AI architecture is built on a multi-layered intelligence stack that combines machine learning, natural language processing (NLP), and generative AI to deliver a truly conversational home loan experience.
Our ML models handle classification, recommendation, and risk scoring, while generative AI enables HOM-i to understand context, intent, and sentiment across text, voice, and even short video interactions.
To ensure inclusivity, we developed a multilingual NLP framework trained on multiple Indian languages and dialects, using a mix of open-source models and proprietary datasets curated from real borrower–agent conversations (with full consent and anonymisation). This allows HOM-i to engage with users in their preferred language and tone.
Data privacy and accuracy remain non-negotiable. All sensitive interactions are encrypted end-to-end, processed through secure, on-premise inference layers wherever required, and continuously benchmarked against bias, hallucination, and compliance parameters.
By combining intelligence, inclusivity and privacy-by-design, HOM-i is redefining how financial services can truly speak India’s many languages.
HOM-i has been trained using 6 lakh call transcripts and over 2 million minutes of advisor interactions. How did your team design this human-AI feedback and training loop? What were some key technical challenges in balancing empathy, context, and precision in decision-making?
The human-AI feedback loop has been central to HOM-i’s evolution. We knew that for an AI assistant to genuinely add value in something as emotional and high-stakes as a home loan, it couldn’t just be accurate; it had to be empathetic and context-aware.
We trained HOM-i using over 6 lakh anonymised call transcripts and 2 million+ minutes of advisor interactions, real conversations between borrowers, agents and lenders. This helped the system learn not just the language of lending but also the language of reassurance, so it could offer clarity and confidence to consumers.
We built a reinforced learning loop with human-in-the-loop validation. Each AI-generated response was rated by our internal advisors on context relevance, emotional tone and factual accuracy. These insights continuously fine-tuned the models, improving both empathy and decisions over time.
Our main challenge was balancing human warmth with data-driven objectivity and accounting for linguistic and cultural differences across regions. We tried to address these with context tagging, emotion classification layers and a safer feedback pipeline that allows HOM-i to keep learning responsibly.
The result is an AI that doesn’t just respond faster; it responds smarter and more humanly, at scale.
You’re targeting 90% precision in loan decision-making within 6–9 months. Could you share insights into the data pipelines, validation metrics, and model governance frameworks implemented to achieve this benchmark while maintaining regulatory compliance?
Achieving 90% precision in AI-led loan decisioning isn’t just about smarter models; it is about strong data discipline and governance. We have built an end-to-end data pipeline that ingests, cleans and standardises inputs in real time, ensuring every data point is accurate, consented and compliant.
Our model validation framework works on three key dimensions, including predictive accuracy, fairness, and explainability. Each output is benchmarked against historical loan performance data, underwriting outcomes, and regional borrower profiles. Continuous drift detection helps us identify and incorporate changes in model behaviour as credit patterns or market conditions evolve.
Governance is at the heart of this system. We’ve implemented a robust Model Risk Management (MRM) layer with strict approval workflows, human oversight, and automated audit trails. Each model version is traceable from dataset origin to production deployment, ensuring accountability and transparency.
To align with regulatory norms, especially those around data privacy, consent, and credit decision transparency, all decision logs are encrypted and accessible through AI interpretability dashboards. This allows both our internal teams and lenders to see the “why” behind each decision, reinforcing trust and compliance.
With ₹30,000 crore in sanctioned loans processed through the BASIC platform, how have you approached scalability, cloud infrastructure, and cybersecurity to ensure uptime, resilience, and trust across your customer and lender network?
Our platform is built on a hybrid cloud infrastructure, leveraging auto-scaling microservices that can dynamically manage traffic spikes, whether from seasonal demand surges or lender-side integrations. This design ensures near-zero downtime and frictionless infrastructure as we onboard new partners across markets.
To maintain uptime and reliability, we use geo-redundant deployments, container orchestration, and real-time monitoring across our cloud environments. Every API call and transaction is tracked through observability layers, enabling proactive anomaly detection and instant recovery protocols.
On cybersecurity, our philosophy has always been ‘trust by design’. All PII/sensitive data in transit and at rest is encrypted, and every user interaction passes through multi-factor authentication and tokenised session management. We also conduct continuous vulnerability assessments and penetration testing in line with RBI and ISO/IEC standards to remain fully compliant.
Finally, we have built a resilience-first culture within our engineering teams from automated rollback mechanisms to 24×7 threat monitoring. In fintech, trust is the real uptime metric. Our goal is to make sure that every transaction processed through BASIC is not only fast and frictionless but also fundamentally secure.
As HOM-i looks to integrate with land records and municipal databases, how do you envision this transforming property verification and loan processing in India? Could this serve as a building block for a larger AI-driven housing ecosystem?
Integrating HOM-i with land records and municipal databases is the next big step towards making home loans truly intelligent and frictionless. Today, property verification is one of the most time-consuming and opaque parts of the lending journey. By connecting HOM-i with verified public and government data sources, we can turn this weeks-long process into a matter of minutes.
This integration will allow HOM-i to automatically validate property ownership, title history, encumbrances, and compliance status, creating a single source of truth for lenders. It not only reduces fraud risk and manual dependency but also improves lender confidence and customer transparency.
This is more than just process efficiency. It’s a foundation for a connected, AI-driven housing ecosystem. When property, borrower, and lender data can interact securely and intelligently, it opens the door for innovations like instant loan sanctioning, digital property passports, and predictive affordability models.
For us, this is a natural evolution of BASIC’s mission to make home ownership simpler, faster, and more inclusive for every Indian. HOM-i’s integration with public data systems could very well become the backbone for India’s next generation of smart, data-driven housing infrastructure.
With India’s continued focus on digital financial inclusion and “Housing for All”, how do you see AI reshaping the lending landscape, particularly in terms of accessibility, decision transparency, and risk assessment?
AI will be the driving force behind India’s next phase of financial inclusion, especially when it comes to housing finance. Where access to formal credit is limited by documentation gaps, digital literacy and linguistic diversity, AI will play an integral role in bridging these divides.
At BASIC, we have witnessed how AI-led automation and regional intelligence make home loans accessible to first-time borrowers, especially in Tier 2 and 3 towns. It is also bringing transparency and fairness to lending. Explainable loan recommendations help borrowers understand why a product or lender suits them best, and multi-dimensional risk models go beyond bureau scores to include alternative data.
AI is transitioning India from credit exclusion to credit inclusion, using intelligent data to make housing finance more transparent, inclusive and equitable.
Looking ahead, what are your top technology priorities and focus areas for 2026? Are there new innovations or product directions, such as predictive underwriting, embedded finance, or advanced GenAI capabilities, that we can expect from BASIC Home Loan?
Looking forward, one of our top priorities is predictive underwriting using behavioural and transactional intelligence to assess borrower intent and repayment potential even before a formal application. This will help lenders move from reactive risk assessment to proactive credit enablement, especially for new-to-credit and informal-segment borrowers.
We are also investing in embedded finance infrastructure, integrating housing-related financial products, from home loans to insurance and renovation credit, directly into partner ecosystems like property portals, developer platforms, and fintech apps.
On the GenAI front, we are building adaptive intelligence systems that learn from real-world interactions, explain their reasoning, and co-pilot human advisors in complex decision-making. Context-aware GenAI copilots for customers and loan officers will bring precision, empathy, and regulatory compliance in every interaction.
Underlying all of this is our commitment to responsible innovation, ensuring every AI advancement aligns with data ethics, inclusivity, and transparency. We want to make BASIC Home Loan the most trusted and intelligent platform, one that not only serves India’s digital borrowers but also empowers the next 100 million homeowners.
