Most offline retailers know the problem all too well: customers walk in, make a purchase, and walk out — leaving behind sales but not a relationship. The real loss isn’t revenue; it’s the data that never gets captured. This is where Artificial Intelligence is rewriting the rules of retail, helping stores turn anonymous footfall into loyal, repeat customers. Leading that revolution is Zithara.AI, a Hyderabad-based retail-tech startup founded in 2020 by Varun Kashyap and Sri Devi Reddy, which has built India’s first ‘customer behaviour platform’ (CBP) to make offline retail as smart, seamless, and data-driven as e-commerce.
Varun Kashyap, co-founder of Zithara.AI, takes CE through their journey of building an AI-driven customer behaviour platform for offline retail, the gaps they identified in traditional systems, challenges of adoption, and the future of retail technology.
Excerpts
Tell us about Zithara.AI?
Sri and I are the co-founders of Zithara. I come from a strategy background at Deloitte, while Sri comes from a retail marketing tech background. She worked for 10 years in the US before moving to India. Both of us always had a love for retail. Retail is one of the biggest segments in the world, contributing nearly $7 trillion globally. Post-COVID, we were contemplating and discussing the future of retail. At that time, the general sentiment was that offline retail — brick-and-mortar — might not exist anymore. But we had a different thought process. We always believed that offline retail was going to grow, and in fact, grow even bigger. While thinking through these things, we realised there were many gaps in how retail was being served by technology. We saw three things happening:
1. Retail was going to grow for sure. Once businesses came back from pandemic, they were entering a different world altogether.
2. Technology gaps were already there but got uncovered more during pandemic. Businesses realised that if they didn’t adopt technology, they would not survive.
3. The rise of AI. Around 2021-2022, AI started entering commercial play, and by 2023 it really started picking up. We were sure it was going to disrupt, so we began researching AI early on.
These three factors became the foundation of Zithara. We incorporated the company in October 2021, actual operations began in 2022, and our first product went to market in 2023. Initially, we were funded by SRiX, an incubator out of Warangal. Being Hyderabad-based, our second round of funding came in January 2024 from HNIs such as the CEO of Darwinbox. That growth capital helped us scale to where we are today.
What are the immediate vulnerabilities in offline retail that make an AI-driven customer behaviour platform necessary?
When we go online, we willingly give away a lot of data — logins, product views, transactions. That’s all first-party data, and technology around it is already well-evolved. In offline retail, there’s also a ton of data — but it’s scattered. When you interact over social media, that data is with marketing agencies. When you enquire in a store, it’s with salespeople. When you buy, it’s in the billing system. Later, when you receive SMS or WhatsApp updates, that’s with communication providers. So the data isn’t absent — it’s disintegrated. For an offline retailer, knowing anyone across all these silos is almost impossible.
And today, offline players aspire to be omnichannel — present online, offline, on apps, everywhere. That makes the challenge even bigger. Large companies like H&M or IKEA have huge tech teams to manage this. But mid-segment retailers — say `500 to `110,000 crore turnover — don’t have the capacity. They rely on external technology providers, who may not fully understand retail. That leads to fragmented implementations. We are at the same point with AI today as we were with the internet in 2004. Back then, you needed servers and modems to get online.
Today, you need structured data to play with AI and most retailers don’t have it. That’s the biggest vulnerability we address.
Customer loyalty programmes and discounts have existed for decades. What gap did you see that pushed you toward building an AI-driven loyalty system?
Traditional loyalty programmes reward only purchases. You buy, you earn points, you redeem — that’s it. But brand loyalty today is broader. For example, you may not buy from a brand but may be talking about it on Instagram. That’s also valuable because you’re spreading information. Brands now want to measure loyalty across the entire customer journey — not just at the purchase stage. That requires a completely different tech stack. Loyalty modernisation was necessary, and to enable that, you need a solid data backbone. That’s what we built.
What has been your biggest bottleneck while introducing CBP to retailers?
Change management — adopting new technology is not easy. If a business is running fine, leaders become complacent. They don’t take drastic decisions until things start going downhill. So our biggest challenge has been educating them. It’s not about the present, it’s about the future. If you don’t adapt now, you risk being stuck in a tough position later. Changing mindsets has been our biggest roadblock.
How supportive has the Hyderabad ecosystem been in your journey?
Very supportive! I’m grateful to be part of the Hyderabad startup ecosystem. We’ve often been asked to move to Bengaluru, but Hyderabad remains home turf and has given us wings. We grew as a company here. Mentors, incubators like SRiX, and organisations have helped us in terms of thought process, growth, and infrastructure. If I had to start this journey again, I would choose Hyderabad again without a doubt.
How do you ensure customer insights remain ethical, secure, and transparent while delivering business value?
First, we are ISO 27001 certified, ensuring data protection. Second, we are DPDP (Digital Personal Data Protection) compliant. That means:
1. Data must be consensual
2. No unethical scraping of data
3. Users should always have the option to delete their data
We strictly follow these compliances and also educate our clients to do the same.
What do retailers usually misunderstand about AI in retail before they see your system in action?
The biggest misunderstanding is that AI magically knows everything about their business, like ChatGPT seems to. But AI is not a silver bullet. It learns only from the data you provide. AI will make mistakes just like humans because that’s how it evolves. Retailers often think, ‘AI will replace humans’. That’s not true. AI will not replace humans, but it will improve efficiency by 30–40%, provided the right data is given. It is here to aid teams, not eliminate them.
What silent risks do you see in offline retail that businesses aren’t paying enough attention to?
1. Rise of organised retail: India is still 6–8% modern retail, compared to places like Dubai which are 98% organised. As organised retail grows, medium retailers who don’t adapt quickly will get wiped out.
2. Rise of e-commerce: Many traditional retailers still believe customers will just walk into their shop regularly. But today, consumers get 200 SMSs and 500 notifications from online players like Zepto, until they buy. Offline retailers who don’t engage with customers personally — through communication, convenience, and fair pricing — will lose them fast. Technology adoption is the only way forward.
Offline retail has always thrived on intuition and relationships. How do you convince retailers that algorithms can read customers better than instincts?
We show them their own data. I sit with retailers and ask them who their best customers are. They’ll name a few familiar faces. Then we run algorithms on their data and the results are completely different. They discover valuable customers they’ve overlooked, people who buy often but never get attention. That’s when they realise that data tells a story that human instinct can miss. If you start listening to the story your own data is telling, the story of your future will change.