are we giving up on Indian AI too soon?
Disclaimer: I quit tech in 2023 because it had become insufferably boring!
Indian companies were trapped in endless cycles of incremental features that nobody wanted to use. The same optimisation requests, the same A/B tests on button colours, the same "innovative" chat bubbles slapped onto existing workflows. Innovation had stagnated into feature factories.
So I did something that the startup community probably considered career suicide: I walked away to meditate.
For 18 months, I disappeared into 10-hour meditation sessions, diving deep into human consciousness and behaviour. I also started teaching yoga and meditation on the weekends to synthesise my learnings. I knew I was being written off by the ecosystem, but I didn't care. I was studying something far more interesting than conversion funnels; the actual mechanics of human decision-making and self transformation.
I started journaling my insights to avoid forgetting the flood of realisations that came during those intense sessions. What began as personal notes became a newsletter called "Meditations" that somehow found 500 subscribers who resonated with my exploration of human condition and consciousness.
At some point, something clicked. The deeper I understood my own mind, the clearer human behaviour became. The patterns I was observing in meditation weren't just personal; they were universal principles of how humans think, resist change, and adopt new behaviours.
The original plan was never to bring this back to work. But watching the incredible potential of Indian AI while sitting on insights about human cognition felt like holding puzzle pieces that could complete a bigger picture.
I find myself energised to work again, but with a completely different lens. My meditation practice has shifted me away from the superficial cognitive level where our busy brains operate to conserve energy. I'm accessing subtler layers of decision-making and pattern recognition that reveal fascinating opportunities.
Maybe all that time wasn't wasted after all.
The Hidden Opportunity in Indian AI
Here's what excites me most about Indian AI right now: we have incredible technical talent building sophisticated products, but we haven't yet fully unlocked the behavioural science that could multiply their impact.
While global AI products like ChatGPT gained massive adoption through intuitive conversational interfaces, Indian AI companies have an opportunity to go even deeper. We can combine our technical excellence with behavioural insights that create products so aligned with human psychology that adoption becomes inevitable rather than effortful.
This isn't about what's wrong with Indian AI—it's about the untapped potential I see everywhere.
We're designing for the wrong competitor
Every AI product team believes they're competing with other AI companies. They're wrong.
Your real competitor is Instagram. And Facebook. And WhatsApp.
Human attention has adapted to 60-second cycles. Our brains have been rewired by social media to expect instant gratification and constant stimulation. When I studied how people interact with technology outside the startup bubble, the pattern was clear: anything that takes longer than 100 seconds to deliver value gets abandoned.
Even Netflix is struggling against Reels and Shorts. Your AI feature better be exceptional if you expect someone to switch from scrolling Instagram to engaging with your product.
This is particularly challenging for Indian AI products. We're still designing for the methodical, patient user who will carefully explore features and gradually understand value propositions. That user doesn't exist anymore.
The Behavioural Economics Revolution We Haven't Applied Yet
During my deep dive into human behaviour, I discovered something fascinating: the most successful products in history weren't just technically superior—they were behavioural aligned too. They understood principles that behavioural economists like Daniel Kahneman and Richard Thaler have been studying for decades.
Let me share what I learned and how it applies to the Indian AI opportunity.
Understanding System 1 vs System 2 Thinking
Kahneman's research reveals that humans operate with two cognitive systems: System 1 (fast, automatic, emotional) and System 2 (slow, deliberate, rational). Most product teams design for System 2 users—rational decision-makers who carefully evaluate features and value propositions.
But real users make most decisions with System 1. They choose based on immediate feelings, familiar patterns, and emotional responses.
This explains why conversational AI succeeded globally. Chat interfaces tap directly into System 1 familiarity—we already know how to text. No System 2 learning required.
The opportunity for Indian AI: design interaction patterns that feel instantly familiar to Indian users' System 1 processing. What are the digital behaviours Indians already perform unconsciously? WhatsApp voice messages, Instagram story interactions, YouTube video scrubbing. AI products that leverage these existing neural pathways will see dramatically higher adoption.
The Power of Choice Architecture
"Nudge" authors Thaler and Sunstein demonstrated that how you present choices matters more than the choices themselves. The same options arranged differently can completely change user behaviour.
Most Indian AI products present users with multiple features and pathways, assuming rational choice-making. But as I observed during my meditation sessions, the mind craves simplicity and clear default paths when cognitively fatigued.
The opportunity: apply choice architecture principles to AI onboarding. Instead of showcasing all capabilities, design a single, obvious default path that delivers value immediately. Make the "right" choice the easiest choice.
For example, instead of presenting new users with 15 AI features, start with one use case that solves their most pressing daily problem then gradually reveal additional capabilities after the first behavioural loop is established.
Creating Variable Reward Schedules
Nir Eyal's "Hooked" model reveals why products become habitual: variable rewards create dopamine loops that keep users engaged. Social media mastered this, but AI products often deliver predictable outcomes.
The meditation insight that transformed my understanding: the human brain is constantly seeking novelty within familiarity. We want reliable results delivered in subtly varied ways.
Indian AI products have a unique opportunity here. Instead of providing identical responses to similar queries, build variability into AI interactions. Same helpful result, delivered with contextual variations that keep the experience fresh. A productivity AI might solve the same scheduling challenge but present solutions in different formats—sometimes as lists, sometimes as visual timelines, sometimes as priority matrices.
Designing for Progressive Engagement
Think about how Progressive Web Apps revolutionised mobile experiences. Traditional websites forced users to download heavy apps or struggle with clunky mobile browsers. PWAs solved this by creating experiences that work beautifully whether you have 30 seconds on a slow connection or 30 minutes on high-speed wifi.
Applied to AI interactions, this means designing for the reality of Indian digital behaviour. When someone opens your AI tool during their morning commute on patchy 4G, they should get immediate value in 60 seconds. But when the same person sits down at their laptop in the evening, the interface should gracefully expand to support deeper, more complex interactions.
For example, an AI writing assistant might offer quick grammar fixes and simple suggestions on mobile during rushed moments, but transform into a comprehensive writing coach with detailed feedback and style analysis when accessed from a desktop during focused work time. Same core AI, different interaction layers based on context and available attention.
This isn't about dumbing down features; it's about smart progressive disclosure that matches human attention patterns.
Leveraging Social Proof and Cultural Context
Indian users make decisions influenced by social context, but not in the way most people assume. Young professionals don't seek family approval for SaaS tools, but they do have distinct behavioural patterns that AI products completely ignore.
Indians are incredibly comfortable code-switching between languages mid-conversation. Most AI products assume monolingual interactions, but Indian users might naturally want to communicate with AI the way they actually think and speak—starting in English, switching to Hindi for emotional emphasis, then back to English for technical terms.
There's also a pattern of sharing interesting discoveries through informal networks. Indians are more likely to share screenshots of useful AI outputs or recommend tools organically. AI products designed with easy sharing and social proof might gain traction through these network effects rather than traditional marketing.
The Investment Phase Creates Commitment
The most sticky products require users to invest something—time, data, preferences—that makes abandoning the product feel costly. This "investment phase" from the Hook model explains why users stick with products they've customised.
Many Indian AI tools miss this entirely. They provide immediate value without creating investment, making users easily replaceable.
The opportunity: design meaningful investment moments. Ask users to teach the AI their preferences, import their data, set personal goals. Not as friction, but as collaborative customisation that makes the AI feel personally valuable.
Understanding Behavioural Change Through Meditation
Through countless hours of meditation and observing my own resistance to change, I learned something humbling: behavioural change is messy, emotional, and deeply personal. It's not about finding the right algorithm or perfect user flow.
During my intense meditation sessions, I watched my mind create elaborate justifications for avoiding beneficial practices. Even when I intellectually understood that longer meditation sessions would help, my brain would generate creative excuses. The breakthrough came when I stopped fighting these patterns and started designing around them; creating conditions where the beneficial choice became the easiest choice.
This personal experience aligns with what neuroscience reveals about behavioural change: the human brain changes at roughly 1.75% per year. Neuroplasticity is real, but it's slow.
During my time studying behavioural transformation, I observed something crucial. Even highly motivated people i.e those attending intensive wellness programs needed multiple exposures and reinforcement before adopting beneficial changes.
If someone struggling with serious health issues needs 5-6 days to embrace life-saving habits, how long do you think it takes for a busy professional to adopt your AI productivity tool?
Most product teams design onboarding as if users will immediately grasp the value proposition. They won't. The average person today is cognitively exhausted, dealing with the attention deficit created by our digital lives, and mentally tired from constant context switching.
Your product needs to deliver promised outcomes within the first 2-3 minutes, not the first week. This is why quick-service apps exploded in India—instant gratification matched the user's mental state.
This is exactly what's missing in most AI product design. We're trying to force behavioural change through logical arguments and feature demonstrations, when real change happens through emotional satisfaction and social reinforcement. The most successful AI products won't convince users to change their behaviour; they'll make the new behaviour feel like the natural next step.
Designing for Present Bias and Loss Aversion
Behavioural economics reveals that humans are present-biased (preferring immediate rewards) and loss-averse (hating to give up what we have). Most AI products promise future benefits that require present effort—a psychologically difficult ask.
What I learned in meditation: the mind readily changes when new behaviors feel like gaining something immediately rather than working toward distant goals.
The opportunity: frame AI adoption as immediate gains rather than future investments. Instead of "This AI will make you more productive over time," try "This AI handles your most annoying daily task starting right now." Instead of "You'll see benefits after a few weeks," design first-session wins that feel instantly valuable.
The path of least resistance always wins
One insight changed how I think about AI adoption entirely: we're designing for people having a rough day.
Not users in perfect conditions with unlimited attention and patience. Real humans dealing with cognitive fatigue, emotional stress, and competing priorities.
The most successful AI implementations I've studied don't fight against these constraints; they embrace them. They assume the user is busy, tired, and skeptical. They design for the least common denominator and win through simplicity, not sophistication.
This is why voice interfaces and chat-based AI gained traction. They reduced friction to nearly zero. No new learning curves, no complex UI patterns, no cognitive overhead.
What this means for AI product strategy
The Indian AI revolution isn't failing because of technical limitations. It's failing because we're optimising for efficiency while humans need familiarity.
We're building features for rational decision-makers while real users operate on emotion and intuition.
We're designing complex workflows while competing for attention spans trained by social media.
The companies that crack AI adoption won't be those with the most advanced algorithms. They'll be those that understand human psychology deeply enough to bridge the gap between technical capability and actual usage.
The Bigger Question About Indian AI
I recently saw Twitter discussions where VCs were advising founders to avoid selling to Indian customers and the heart breaking slack received by Sarvam AI for its reportedly low user engagement numbers. This made me wonder: are we giving up on Indian AI too quickly?
What if the issue isn't that Indians don't like Indian AI, but that we're designing AI products using assumptions that don't fit human behavioural patterns?
The Collaborative Opportunity Ahead
After 18 months of studying human consciousness and behaviour along with a decade spent on product strategy, I see an incredible opportunity to collaborate with Indian AI founders who want to multiply their technical capabilities with behavioural science insights.
This isn't about fixing what's broken, it's about amplifying what's already working. Taking technically excellent AI products and designing human interfaces that make adoption feel natural, engagement feel rewarding, and retention feel inevitable. UI for AI has lot of potential beyond the now popularised chat interface.
The companies that get this right won't just succeed in India. They'll create behavioural frameworks that demonstrate how deep human understanding can unlock AI's potential anywhere in the world.
If you're building AI products and are excited about exploring how behavioural science could amplify your technical capabilities, if you're curious about designing for actual human psychology rather than ideal user scenarios, I'd love to collaborate and learn together (either by investing or helping with product strategy, or both).
Signing off,
Kalyani Khona
I write about stuff like this in my newsletter The Third Frontier so consider subscribing to it 👇
