In an exclusive conversation with CIO&Leader, IndiaMART’s CIO Nikhil Prabhakar engineered a future-ready AI roadmap, one that balances business impact, security, and continuous innovation to transform customer experience at scale.
CIO&Leader: How do you define success when transitioning an AI initiative from pilot to production?
Nikhil Prabhakar: Improvement in CX (customer experience) remains a top priority for us, while implementing any initiative/solution within the organisation. We identify the customer problem, based on their feedback and journey on the platform, simplify those issues to see how they can be resolved, and re-engineer the existing processes using technology to better serve the customer.
Additionally, our decision framework takes into account security and innovation as the two fundamentals for any solution. These also remain our KPIs to measure the success of every initiative, from pilot to production.
● Business Impact First: Post implementation, we assess whether the AI initiative is enhancing user experience, operational efficiency, or marketplace trust. For instance, when AI improves lead quality, automates support interactions, or delivers personalized experiences, it moves to the top, highlighting that this is working in favor of the customers.
● Security by Design: Another area to measure the success is the security architecture and how the solution responds to checks such as data privacy, control, explainability, and ethical usage.
CIO&Leader: What are the core pillars of your current enterprise AI strategy?
Nikhil Prabhakar: IndiaMART is one of the first internet companies of India, established in 1996. From embracing the internet to integrating AI for more than 8 years now, our strong foundation and tech maturity allow us to confidently scale more advanced initiatives today. We started using AI in classical machine learning and NLP models across areas like search ranking, fraud detection, and categorization, and are now integrating it into deeper workflows such as auto-categorization and buyer-seller matchmaking.
● Modernisation of applications: Our AI-ready workforce is focused on modernisation of applications via containerization through Kubernetes & Docker. This enables seamless orchestration of containerized applications, efficient management of microservices architectures, and ensures rapid deployment. Our transition to service-oriented architecture (SOA), multi-cloud, and Kubernetes has enabled flexible experimentation and deployment of AI models at scale. Additionally, we keep expanding on the consolidation of Logic and the promotion of functional programming, embracing useEffect. These functional components promote stateless components, fostering cleaner architecture, thus modernising the code base further.
● Continuous improvement: We are actively building and fine-tuning open-source foundation models on our proprietary datasets. This includes image-based product identification, data cleaning, and smart assistance layers. Our focus is not just on adopting GenAI but aligning it deeply with specific business workflows. Data is our competitive moat, and we don’t solely rely on generic AI models. Instead, we fine-tune open-source models on IndiaMART’s rich proprietary datasets—text, images, voice, and transactional data—to tailor them to our specific business context.
● Use Case Expansion: From customer experience to sales productivity, finance operations to HR onboarding processes, we are embedding AI across support functions, not just customer-facing products or core matchmaking platforms. Over time, we will be able to reduce costs, increase development speed, and enhance system reliability by embracing these open-source solutions.
CIO&Leader: What key AI use cases have successfully moved into production, and what measurable impact have they delivered?
Nikhil Prabhakar:Our initiative to fuel AI-Powered Conversational Commerce has not only streamlined business transactions but also reinforced trust in digital B2B commerce. The introduction of IM Insta – a seamless integration of WhatsApp with the Lead management system – is a key initiative to help sellers instantly connect with buyers. This has helped in reducingfollow-ups, ensured less intrusion, and increased buyer responsiveness by 3x. Similarly, our new photoSearch 3.0 model also reduced average response time searching products on the platform from 4.7 seconds to just 2.5 seconds and enhanced the model’s accuracy significantly over traditional models. This has ultimately improved the relevance of the search results and reduced the margin of error, allowing us to provide the most pertinent results to our users’ queries.
Additionally, Generative AI (GenAI) and Agentic AI are two transformative technologies delivering tangible ROI today. With Gen AI integration, we are fine-tuning large language models using proprietary data and embedding them within our ensemble. It combines the strengths of both the rule-based engine and machine learning model, leading to a more precise and effective matching process. This has helped us in achieving efficiency internally and reaching around 90% accuracy in automated product classification.
CIO&Leader: What kind of infrastructure, organisational and workforce-related challenges you have faced scale AI effectively within your organization?
Nikhil Prabhakar: Scalability is integral to our operations, and increased use of AI within the organisation is not a challenge, but an opportunity for us. This not only includes architectural changes, but also skilling the workforce and ensuring its future-readiness to embrace innovations at a faster pace. A couple of areas where we continue to work for faster yet responsible scalability are:
● Data Quality & Readiness: Ensuring high-quality, vast, and consistently labeled data for training and real-time inference via robust data governance frameworks and automated data validation pipelines. This helps us to maintain integrity across our massive buyer-seller interaction datasets.
● Seamless Integration into Core Workflows: Our microservices architecture and flexible APIs have been crucial in allowing incremental integration and enabling AI to power features like instant lead connections and enhanced search, thus moving AI from pilots to deeply embedded solutions.
● Model Governance & Trust: We also keep working on ensuring that our AI systems are ethical, explainable, and free from biases or ‘hallucinations’. Particularly with Generative AI integration in critical functions like lead verification and communication, our focus remains on establishing a comprehensive internal AI Governance Framework that includes ethical standards, usage boundaries, and continuous audit mechanisms.
● Deep Convergence of AI & Security: Instead of reactive, rule-based defenses, we’re transitioning to AI as our first line of cyber defense. This means deploying AI models that proactively learn evolving attack patterns, detect subtle signals of abuse, and auto-mitigate threats (e.g., spam, fraud) before they impact users or lead to data leakage.
● Comprehensive cybersecurity training: To ensure our workforce remains vigilant, we have clear usage Guidelines for Public AI Tools such as ChatGPT, Copilot, or code generators in place to ensure what data can and cannot be shared with third-party platforms.
● Upskilling and reskilling: Owing to an AI-skilled workforce and the best tech talent working with us, we have been developing AI-based engines for matchmaking and search for almost 8+ years now. We also try to train the teams for upskilling, conduct weekly sessions, and encourage them to pick up one small use case at a time and solve it quickly, efficiently with AI. Internally, we conduct hackathons to support IndiaMARTians and provide them use cases to experiment and learn AI.
In essence, we’re building a future where AI not only drives business value but also secures the platform, and the platform, in turn, secures the AI, redefining risk and trust in our digital ecosystem.
CIO&Leader: Looking ahead, what does your AI roadmap over the next 12–18 months look like – especially in terms of GenAI or foundation model deployments?
Nikhil Prabhakar: Our focus remains on enhancing Customer Experience, driving efficiency, and maximizing ROI for businesses on our platform with the help of technology and increased AI integration. Over time, we continue to innovate on their behalf and provide them with the best possible solutions. Our focus areas remain:
● Better CX: Ensure hyper-personalized, context-aware responses and resolve complex multi-turn queries instantly. In addition to this, we will integrate AI to provide proactive assistance, thus reducing resolution times dramatically.
● Efficient product/catalog management: With time, our systems will intelligently infer product details from minimal inputs (e.g., a single image or brief text), create & map the right attributes, and categorizations. This will ensure faster and easier listing of products on the website, while also ensuring they are error-free.
● Analysis: Use of AI to understand market trends, demand forecasts, pricing strategies, competitor analysis, and receive immediate, actionable recommendations. This not only helps in supporting the customers better but also helps sellers grow their business better on their own.