How Startups Deal With Data Challenges In The Race To Scale Artificial Intelligence  

As we hit the midpoint of 2025, one thing is evident. Gen AI (generative artificial intelligence) is no longer a promising experiment or an emerging landscape. It is a full-fledged transformation engine. From workflow automation to software development, GenAI is redefining how businesses operate, making systems faster, smarter and more adaptive. But the real disruption lies beneath the surface, determined by how organisations collect, manage and scale their data.

Legacy systems — rigid schemas, siloed data repositories and batch-based analytics — are cracking under the pressure of mission-critical requirements. Replacing them is a new data backbone built on real-time streaming, cloud-native platforms (for access and scale), and autonomous databases. These modern stacks are critical to unlocking the full potential of GenAI, powering everything from fraud detection and customer segmentation to hyper-personalised recommendations.

The numbers back the momentum. GenAI analytics is expected to surge from $1.7 Bn in 2025 to $5 Bn by 2029, growing at an impressive 31% CAGR. Meanwhile, the AI infrastructure market is projected to reach $394.5 Bn by 2030, up from $135.8 Bn in 2024 at a 19.4% CAGR. While India-specific data remains sparse, APAC markets will accelerate rapidly.

For Indian startups, this signals a strategic shift. They are no longer just tweaking systems but building from the ground up. The GenAI era is pushing them into continuous learning loops where innovation is not an add-on. It is baked into the core of what they build.

To unpack how high-growth startups are re-engineering their data stacks to enable seamless AI execution, Inc42 and database software developer Couchbase co-hosted a closed-door roundtable in New Delhi — the first edition of Boardroom: The AI & Data Playbook.

The session brought together nine tech leaders from various sectors, including foodtech, medtech, ecommerce, cloud, AI and more, to discuss the challenges and strategies around building AI-ready infrastructure.

The conversation, moderated by Ratnaafin Capital’s CTO Kshitij Shah, explored the core theme: How Startups Deal With Data Challenges In The Race To Scale Artificial Intelligence. The participants included:

• Dr Shakti Goel, chief architect & data scientist, Yatra Online
• Gaurav Bagga, SVP and head of engineering & product, Pristyn Care
• Shantanu Prakash, director (data analytics), CashKaro.com
• Pankaj Judge, chief for enterprise SaaS, EV and sustainability, Droom
• Sunil Rai, senior director (tech), Magicpin
• Diwas Sharma, tech head, Fashinza
• Rahul Prasad, cofounder & CTO, Bobble AI Technologies
• Amit Gupta, cofounder & CTO, TrulyMadly
• Krishna T., regional business head, Couchbase

AI-Powered User Profiling: Turning Data Into Deep Personalisation

During the roundtable, speakers underscored how user profiling has matured beyond basic demographic segmentation to granular, personalised insights. Today, AI systems sift through browsing history, purchase behaviour, geolocation data, social media activity and interaction logs to identify nuanced behavioural patterns. The outcome is hyper-personalised user experiences that convert data into measurable outcomes, such as right-pricing pre-owned vehicles or matching customers with the right vehicles. 

Droom, an online marketplace for buying and selling new and pre-owned vehicles, has developed Orange Book Value (OPV), a cornerstone of the platform’s data-driven strategy. According to Pankaj Judge, OPV uses more than 78 parameters such as location, vehicle variant, fuel type, segment, city-specific RTO norms and even GST fluctuations, to determine the fair market value of a used vehicle. 

“Say you purchased a car in 2023 and want to sell it in 2025. Of course, the price will depend greatly on the make, model and trim level. But we also factor in city-based depreciation, local taxes, total distance travelled and other critical parameters,” he explained. “For instance, cars in New Delhi typically hold value longer than those in Kolkata or Indore. We run these variables through complex statistical models like Monte Carlo regression to predict depreciation curves. This analytical muscle powers Droom’s broader personalisation tools and results in highly accurate, location-sensitive valuations.” 

Its data-driven recommendation engine, Droom Suggest, also provides the most suitable buying options based on user inputs such as budget, passenger capacity and use case (personal or commercial transportation, daily commute, occasional/long-distance travel or multipurpose usage) to ensure a seamless transactional experience.

TrulyMadly, an online dating and matchmaking platform, has tapped into GenAI and large language models (LLMs) to supercharge user profiling beyond the standard data in sign-up fields.

“Users typically enter basic information — education, profession, likes and dislikes, and upload a profile photo. Now, GenAI helps us enrich those profiles in ways that were not possible even two years ago,” said cofounder and CTO Amit Gupta.

“The platform uses artificial intelligence to extract contextual cues from images. A selfie at the Eiffel Tower may signal a love for travel; a shot on a Goan beach may reveal a taste for parties or coastal life; a photo in the mountains? Someone is likely into nature or adventure. It is about decoding deeper personality traits to drive more meaningful matches,” he explained.

The only glitch: What will happen if these images are photoshopped and send the wrong signals? Fortunately, AI tools are being developed to detect manipulated imagery. TrulyMadly is also trying to address this pain point, and its current accuracy rate is around 98%.    

How Startups Are Using GenAI To Improve Operations & Drive Growth

Startups today are rapidly implementing GenAI to streamline operations, boost efficiency and deliver tailored customer experiences. GenAI tools automate time-consuming and resource-heavy manual workflows like staff training and customer engagement with measurable impact by harnessing machine learning (ML) and natural language processing (NLP).

Healthtech brand Pristyn Care, which makes elective surgery accessible and affordable, is a case in point, as Gaurav Bagga shared how GenAI improved its day-to-day operations. A primary use case was training care co-ordinators.

“We have 500 to 600 of them who interact with patients daily. But given the routine churn and ongoing hiring, manual training was not scalable,” he said.

So, the startup built a GenAI-based training platform featuring a human-like avatar that delivers all standardised modules. After the training, co-ordinators are evaluated through GenAI-based assessments.

“Besides training, we have built a Copilot-like system to help our care co-ordinators interact with patients,” said Bagga. “When we rolled out the system 18 months ago, its accuracy in identifying high-potential leads was around 67%. Today, it exceeds 90%. It also recommends timely follow-ups based on past conversations and sends automated reminders, helping co-ordinators stay organised and proactive.”

As GenAI Advances, Traditional Roles & Industry Dynamics Get Rewired

The rise of GenAI is redefining traditional roles across industries. As GenAI tools become embedded in daily operations, they reshape how teams collaborate, automate workflows and tackle challenges, from speeding up production pipelines to scaling up customer support.

AI tools make tasks easier and less time-consuming to help people engage more meaningfully with technology, strategy and creative problem-solving. Although GenAI can accelerate workflows and boost productivity, it does not replace domain-specific human expertise. In fact, this gets amplified.

Fashinza, a B2B startup diving deep into manufacturing after pivoting from a marketplace for fashion procurement, said that the role of product managers changed following GenAI adoption.

“Six months ago, we had dedicated product managers who acted as the bridge between business requirements and technical execution. Today, there is no such role in the organisation. All product requirements now come directly to me, and we handle them internally, aiming to build tech-driven use cases,” said technology head Diwas Sharma. 

A critical part of working with AI is knowing what to ask — a skill that requires domain knowledge and a good understanding of GenAI to frame the right problems and interpret responses meaningfully. 

“If you can’t pose the right questions, you can’t build the product you need,” he said. “These roles are evolving. They are no longer limited to defining a product vision. One must ensure that AI tools align with that vision to deliver value. And the bar is only getting higher.”

Traditional product leads, he observed, would often struggle to adapt. “Many still propose interfaces with standard input fields, generating linear approaches. But we are going for more advanced approaches — like letting users upload raw email messages — from which the system automatically extracts relevant information.”

Krishna T. from Couchbase concurred, emphasising the changing dynamics across industries traditionally relying on large workforces.

“Consider the IT services sector. Giants like TCS and Infosys are hiring fewer people, not because business is bad. It is because GenAI and automation drive greater efficiency. As these technologies scale, companies will shift to leaner, more profitable operating models, even as headcounts decline.”

The broader trend will hold. Industries will evolve; old roles will disappear; new opportunities will emerge and streamlined operations will lead to bigger margins, even with fewer people on the payroll.

Choosing Cloud Infrastructure Is Critical For Startups 

Startups keen to opt for cloud infrastructure face a trade-off between control and convenience. However, Rahul Prasad, from the conversational AI platform Bobble AI, urged founders to prioritise speed and convenience by leveraging managed services early on. As startups mature and scale, cost optimisation and vendor lock-in will become critical considerations.

Shantanu Prakash, from the cashback platform CashKaro, emphasised that no single strategy could serve all. Larger organisations often buy off-the-shelf solutions to speed up time-to-market, while startups may build the entire tech stack in-house to differentiate themselves. Yatra Online’s Dr Shakti Goel also detailed why the online travel agency (OTA) shifted from an on-premise setup to the cloud, driven by the need for scalability and modernisation.

That is a valid point. 

High-growth startups are increasingly turning to AI-ready data stacks and cloud infrastructure to unlock the full potential of AI-driven innovation. Over the ages, the ability to stay agile and cater to new opportunities has been the key to business success. (Think Apple, and we need not go any further.) But as the roundtable unanimously concluded, regardless of disruptive technologies and the speed of adoption, long-term growth and profitability remain the endgame.