Akshay Mani Aggarwal

My Team at CheQ no longer writes a PRD. #Wisor our #Credit #Card #Advisor is a living example of the same. We use Claude Code/Lovable to build the first version of the feature themselves. Takes the feedback, gets the approval and then writes a detailed PRD for engg with the working prototype to understand. We have been following this for a while, and the results are simple. 1. Faster Approvals with clear understanding of how it will look and work when it goes to Prod. 2. Less your visual thinking vs my visual thinking. 3. No back and forth on Iterations by design and Engg. This is what happens when you eliminate 𝗛𝗮𝗻𝗱𝗼𝗳𝗳 𝗗𝗿𝗶𝗳𝘁. In a traditional org, the cycle kills you: > PM writes a spec. > Design interprets the spec. > Engineering interprets the design. > Each handoff introduces drift. Each handoff adds a week. The standard 4-week cycle (PRD → Design → Ticket → Sprint) compresses to 𝗱𝗮𝘆𝘀 when the PM can "Vibe Code" and product results. The math is simple. If your experimentation cycle is 8 weeks and your competitor's is 3 days, they are running 13𝘅 𝗺𝗼𝗿𝗲 𝗲𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁𝘀 per quarter. Compound that learning gap over a year, and it becomes insurmountable. You choose what works for you!! Did you try #Wisor as of now? Aditya Menon

88

10 Comments

It’s becoming harder to justify a product manager who can’t vibe code. I regularly prototype full functionality when communicating requirements to my engineering stakeholders. Tools like Cursor, Bolt, and Claude allow me to go beyond words and show what’s possible. I now even push code to GitHub, giving my engineering partners visibility into what’s under the hood. These prototypes aren’t final designs, they’re agile artifacts that provide higher context and clearer solution intent. Vibe coding empowers product managers to move faster, give engineers clarity, and ultimately spend more time with customers. And that’s always a win for products, users, and builders alike. #ProductManagement #VibeCoding #AI #Engineering #Prototyping #Agile #Leadership #ProductDesign #Innovation

11

8 Comments

Why Bank Integrations Pass SIT but Fail in UAT One thing I’ve repeatedly seen in fintech projects: Bank integrations pass SIT… and then fail in UAT. The confusing part? The code works. APIs respond. Transactions are successful in testing. Yet go-live gets delayed. The reason is simple — SIT tests technology. UAT tests operations. In SIT we validate: • request/response format • authentication • API connectivity • success and failure codes But in UAT the bank checks a completely different world: Did settlement happen in the correct cycle? Are reversals handled the same day or next day? Does reconciliation match the bank statement? What happens on cut-off time? How are holidays handled? What if a transaction is successful but acknowledgement is delayed? This is where many integrations break. Because integrations don’t fail at APIs. They fail at process assumptions. I’ve seen cases where: – payment was successful but not settled – settlement happened but report format differed – transaction worked but reconciliation mismatched – retry logic created duplicate entries From a development view the system worked. From a bank’s operational view it was not usable. The learning for me: A bank integration is not a technical integration. It is an operational integration. You are not only connecting systems. You are aligning two financial processes. The earlier teams understand settlement, reconciliation, cut-offs and exception handling — the smoother UAT becomes. Otherwise UAT becomes the phase where the real requirements are discovered. And that is why many fintech timelines slip right before go-live. #Fintech #Payments #BankingTechnology #SoftwareDelivery #SystemIntegration

25

1 Comment

𝐏𝐞𝐫𝐩𝐥𝐞𝐱𝐢𝐭𝐲 just made a bold move—offering a 12-month Perplexity Pro subscription (worth ₹17,000/year) for free to its entire Indian user base, potentially reaching 360 million people. Let that sink in. That’s ₹𝟲.𝟭𝟮 𝗹𝗮𝗸𝗵 𝗰𝗿𝗼𝗿𝗲 (~$𝟳𝟯 𝗯𝗶𝗹𝗹𝗶𝗼𝗻) worth of product being distributed—at zero direct cost to users. So, what’s the play here? 𝑰𝒔 𝑷𝒆𝒓𝒑𝒍𝒆𝒙𝒊𝒕𝒚 𝒇𝒐𝒐𝒕𝒊𝒏𝒈 𝒕𝒉𝒆 𝒃𝒊𝒍𝒍? Or is this a smart growth hack—a strategic marketing bet that saves millions in user acquisition? Having worked with 𝙒𝙮𝙣𝙠 𝙈𝙪𝙨𝙞𝙘 and 𝑨𝙞𝒓𝙩𝒆𝙡 𝙈𝒐𝙫𝒊𝙚𝒔, I’ve seen firsthand how telcos can drive massive app adoption by bundling services into their plans. App installs that would cost ₹200–₹300 each via ads can be acquired at a fraction of the cost through native ecosystem pushes. 𝑨𝙣𝒅 𝒉𝙚𝒓𝙚’𝙨 𝙩𝒉𝙚 𝙠𝒊𝙘𝒌𝙚𝒓: when telcos send targeted SMSes and push notifications, the install rates skyrocket. This kind of direct distribution—at scale—is nearly impossible for an app like Perplexity to achieve on its own. This might be Perplexity’s moment—𝙟𝙪𝙨𝙩 𝙡𝙞𝙠𝙚 𝙐𝙗𝙚𝙧’𝙨 𝙚𝙖𝙧𝙡𝙮 𝙙𝙖𝙮𝙨 in India, when they subsidized rides and drivers to build irreversible habit loops. Today, millions can’t imagine life without it. And with 𝗣𝗲𝗿𝗽𝗹𝗲𝘅𝗶𝘁𝘆'𝘀 𝗻𝗲𝘄 𝗯𝗿𝗼𝘄𝘀𝗲𝗿 launch, this partnership could also be laying the groundwork for a full-stack ecosystem: AI assistant + native browser + telco distribution. 🚀 𝐀 𝐰𝐢𝐧-𝐰𝐢𝐧: --Airtel gets to showcase cutting-edge AI to its users. --Perplexity scales faster and cheaper than traditional routes. --Users gain premium access—no strings attached. 𝑾𝒉𝒂𝒕 𝒅𝒐 𝒚𝒐𝒖 𝒕𝒉𝒊𝒏𝒌—𝒔𝒎𝒂𝒓𝒕 𝒑𝒂𝒓𝒕𝒏𝒆𝒓𝒔𝒉𝒊𝒑 𝒐𝒓 𝒓𝒊𝒔𝒌𝒚 𝒈𝒊𝒗𝒆𝒂𝒘𝒂𝒚? #AI #Perplexity #Airtel #ProductGrowth #TelcoStrategy #AppDistribution #UserAcquisition #Wynk #BrowserWars #FreeBees

16

India just redefined AI-powered commerce. And it's not about chatbots. It's about trust, scale, and guardrails. On Oct 10 at Global Fintech Fest, NPCI, OpenAI, and Razorpay announced a collaboration that could change how 500M+ Indians discover and pay for products. But here's what excites me as someone building AI systems: This isn't just about AI agents processing payments. It's about building AI commerce infrastructure with: ✅ Compliance-first design - Built on India's trusted UPI rails ✅ Conversational commerce - Shop in your language, pay instantly ✅ Responsible AI at scale - Guardrails embedded from day one ✅ Financial inclusion - Bringing AI benefits to tier 2/3 cities The questions I'm asking: 🔹 How will AI agents verify user intent for high-value transactions? 🔹 What guardrails prevent fraud in conversational checkout flows? 🔹 How do we ensure AI doesn't push unnecessary purchases? 🔹 Can this work seamlessly across 22+ Indian languages? 🔹 What's the compliance framework for AI-driven financial decisions? Why this matters: We've seen AI content generation scale globally. But AI-powered commerce at India's scale—with built-in payment infrastructure and regulatory alignment—this is uncharted territory. If done right, this could be the blueprint for responsible AI commerce worldwide. My hope: That we build this with the same rigor we built UPI—user-centric, secure, and accessible to every Indian, not just the tech-savvy few. Kudos to Pragya Misra, Dilip Asbe, and the teams at National Payments Corporation Of India (NPCI), OpenAI, and Razorpay for taking this bold step. To fellow AI builders: How do we ensure AI commerce agents are helpful, not manipulative? Where should the guardrails be? This is the conversation India—and the world—needs to have. #AICommerce #ResponsibleAI #UPI #FintechInnovation #IndiaStack #AICompliance

6

My notes from Razorpay #FTX26 🚀 Spent yesterday at the Lalit Ashok for #FTX26. It’s clear the industry is moving past simple APIs and into what Razorpay is calling the "Agentic Era." As an engineering leader, these were the real highlights for me: 1. The "2-Minute" KYC: Seeing onboarding go from days to just 2 minutes using Agentic KYC was impressive. It’s a great example of using AI to fix a massive backend bottleneck rather than just adding a "chat" button. 2. The Agent Studio: They’ve built a way to deploy "digital employees" for things like money collection, negotiations, and dispute resolution. It shifts the dev focus from writing rigid logic to building "reasoning" rails. 3. The Infrastructure Layer: Caught some solid discussions on the "Ground 0" problems—chips (Nvidia), energy, and the massive compute needed to make these agents reliable at scale. 4. Multiple candid discussions(in parallel) with industry leaders on what the next decade looks like in the Agentic Era. Especially, building business abstractions for businesses. The big takeaway: The challenge for us as engineering leaders is changing. It’s no longer just "can we scale this?" It’s "Can we trust this system to reason correctly?" Kudos to the Razorpay team for a power-packed day. India’s FinTech roadmap just got a lot more interesting. #FTX26 #Engineering #FinTech #AI #Razorpay #SystemDesign

30

Finance teams are evolving faster than most people realise. What used to be reporting and reconciliation is becoming automation, intelligent systems and AI-driven decision layers. AI is starting to reshape how finance operates. At Razorpay FTX26, we’re opening up the Circle of Expertise - an open space where builders, operators, can discuss how financial infrastructure is changing and what the next phase might look like. I’ll be talking about the rise of AI-powered finance and if you’re building or running finance systems and have a question about the future of finance, send it through, I’ll answer it live! Drop it here → https://lnkd.in/gEU_9sNF The LaLit Ashok 🗓️12th March, 2026 https://lnkd.in/gcRM3x8K

140

1 Comment

It was indeed an enriching experience being a panelist at "AI in Testing & Testing AI", hosted by Pradeep Soundararajan's team at Moolya – an event that was not just a gathering, but a bold, no-fluff exchange of ideas on how AI is actively transforming the future of Quality Engineering. Some key reflections from the evening:- -> AI isn’t replacing testers — it’s expanding our capabilities and providing additional toolkits. From intelligent test generation to agent-driven automation & exploratory testing, we’re entering a new era of augmented quality and this is just the beginning. -> Testing AI applications and models requires a different mindset & specialised skills - Traditional test strategies fall short, we need strategies that are ethical, bias-aware, and deeply contextualised to assess the non-deterministic nature of LLMs and AI systems. -> AI in testing is not just about automation or speed - It’s about adaptability, understanding context, and enabling more informed, data-driven decisions throughout the lifecycle. -> And I believe the most important insight: human judgment remains irreplaceable - The future of testing isn't human vs AI — it's Human with AI. What I enjoyed the most?  -> The raw, unfiltered curiosity from the such a large audience through live polling. -> The diverse & insightful perspectives from co-panelists — thank you Sathish K S, Nilanjan Dutta, Sharad Singla, and Amit Vyas for steering such an engaging conversation. #LeadershipInTech #AIinTesting #TestingAI #GenAI #QualityEngineering #MoolyaEvents #QAOnTheRocks #SoftwareTesting #AgenticAI #BangaloreTech #FutureOfTesting #TestingCommunity

19

2 Comments

Prakhar's HLD for ~20,000,000 daily file uploads (part 2) let's see how that video appears on phone 1. Sync Signal 1. Phone's Dropbox client receives the sync notification 2. It checks with the Metadata Service for details about the new video 2. Metadata First 1. Client downloads the video's metadata first 2. Client see the video file in the folder list almost instantly, even before the content is there 3. On-Demand Download 1. When we tap to play the video, the client requests the block locations 4. Secure Block Retrieval 1. Block Service provides secure links for the phone to download each block 5. Parallel Download & Reassembly 1. Client downloads the blocks in parallel 2. it reassembles them into the original video file on the device 🤔 over to you, what are benefits of on-demand Download? Let me know in the comments! 👇 📣 Join 7-day email course to effectively optimize 60-minute system design interview - https://lnkd.in/gDBzCnE3 

62

23 Comments

I just published a detailed deep dive on designing a Commerce Intelligence Platform for India’s fragmented quick-commerce ecosystem. Brands today sell across Zepto, Blinkit, Flipkart Minutes, Amazon — but struggle to answer: • What is my true category market size? • What is my brand share vs competitors? • Why did my organic rank drop? • Which new launches are gaining momentum? This article covers: ✔ Full database schema (entity → fact → mart layers) ✔ Rank-based market size estimation using Zipf’s Law ✔ Sponsored vs organic separation logic ✔ Tier-based accuracy system (100% accuracy for top SKUs) ✔ Microservices architecture ✔ LLM integration for classification, entity resolution & root-cause analysis ✔ SQL queries for every major business insight It’s a complete master architecture — end to end. Would love feedback from product, data, and commerce folks. #Ecommerce #QuickCommerce #SystemDesign #DataEngineering #AI #ProductArchitecture

3

Immensely proud to see the work we do at Dezerv featured in Season 2 of the Bharat Fintech Story, now streaming on JioHotstar. For us in the engineering team at Dezerv, this is a special moment. It shines a light on something we have been obsessed with from day one: building an institution on the bedrock of technology, trust, and security. As I mention in the episode, in wealth management, trust isn't just a regulatory checkmark; it's the entire foundation. That’s why we engineered Dezerv from the ground up with a tech-led approach. Technology is the enabler that allows us to deliver our expert, data-driven investment philosophy at scale, but it’s the trust we build through that technology that truly matters. This feature is a testament to the incredible work of our team members at Dezerv. A huge thank you to them, to our clients who place their faith in our platform, and to Edstead and JioHotstar for telling our story. I invite you to watch the episode to see how we're building the future of wealth-tech in India. Watch it here: https://lnkd.in/g7Y4nDe4 Sonal Singh Kanav Gupta Arijit Mukhopadhyay Mayank Kumar Tauseef Perwez Nitya Tiwari Eshaan Ramnani Satyam Singh #Dezerv #WealthManagement #JioHotstar #BharatFintechStory

177

4 Comments

The cost of convenience is most often privacy. More than money itself. Eg. Identity - Aadhaar - Surveillance in wrong hands Travel - DigiYatra - Private company and surveillance possibilities Health - ABHA - Medical record sharing Financial Data - Account Aggregators - Data leakage and concentration Why do you need centralisation of everything? It all leads to only 2 things - surveillance (in the name of trust) and taxation (to add to the burden of common man). The funny thing is, people often don’t really understand what privacy is. Then fall for digital arrest, cyber scams, surveillance like Pegasus spyware, etc.

35

7 Comments

PayPal Joins Hands with NPCI to Take India’s UPI Global 🌍 India’s homegrown payment system is officially stepping onto the world stage. In a major leap for cross-border digital payments, PayPal has partnered with NPCI International Payments Limited (NIPL) to bring India’s Unified Payments Interface (UPI) to PayPal World—PayPal’s newly launched global payments platform. Announced in July 2025, PayPal World connects some of the world’s largest digital wallets under one ecosystem—PayPal, Venmo (US), Tenpay Global (China), Mercado Pago (Latin America), and now UPI (India)—together representing nearly 2 billion users worldwide. So what does this mean in real life? 👉 Indian users can now pay international merchants using UPI, starting with select PayPal merchants in the US. No cards. No confusing forex steps. No new apps. Just the same UPI experience—fast, familiar, and real-time—now working beyond India’s borders. Ritesh Shukla, MD & CEO of NIPL, summed it up well: “This integration marks a major milestone in expanding UPI’s global footprint and aligns with our vision of seamless, secure, and inclusive cross-border payments.” For global merchants, this is huge. It opens access to hundreds of millions of UPI users in India, reduces checkout friction, and makes cross-border e-commerce more inclusive. PayPal CEO Alex Chriss has also highlighted that UPI is the first payment system partner on PayPal World—a strong signal of India’s leadership in real-time digital payments and its influence on global standards. This isn’t just a technical integration. It’s a statement. India’s fintech innovation—once focused on domestic payments—is now exporting simplicity, speed, and scale to the global economy. 🚀 Will this accelerate global adoption of UPI-like systems? 💸 How do you see this impacting cross-border e-commerce and remittances? The future of global payments just got a little more Indian 🇮🇳 #UPI #FinTech #DigitalPayments #IndiaTech #CrossBorderPayments #PayPal #paytm

84

System Design Part 2 Continuing my System Design Blog Series, this post is a deep dive into the Low-Level Design of WhatsApp. 📱 From handling 2 billion daily messages to ensuring high availability, we’ll walk step by step through how WhatsApp is architected at scale. 🚀 High-level estimate Here is our high-level estimate: TypeEstimateDaily active users: (DAU)50 million Requests per second (RPS) : 24K/s Storage (per day) : ~10.2 TB Storage (10 years) : ~38 PB Bandwidth : ~120 MB/s

6

Go Performance Tuning at Scale: Zepto’s pprof Journey (Part 1) At Zepto, performance isn’t just a priority — it’s an obsession. The best part about making systems more efficient is that it not only boosts speed but also drives real cost savings. In this blog, Akash Agarwal shares our journey of performance tuning — the challenges we tackled, the wins we unlocked, and how the process didn’t just improve our systems, but also shaped the way our engineers think about building scalable, efficient systems. Blog link: https://lnkd.in/gtzdtmGg #ZeptoTechXpress #PerformanceTuning Shivanshu Kumar Tiwari Ashish Bhatt Sumit Gupta Aditya Haridas Menon Harshal Khodifad Abhinav Nimesh Uday Solanki Devansh Chaudhary Devansh Goswami Sumit Ojha Vivek Gover Prashant Giri Balram Vishwakarma Shubham Prashar Anoop Singh Mayank Agarwal Nikhil Mittal

124

If ONDC made sense, Agentic commerce absolutely will kill it. Stripe launched Agentic commerce protocol for US, while the infra and market is ready in India. An opportunity for Indian Fintech space. The whole value ecosystem of marketplaces to connect buyer and Seller via Search, Algorithms and recommendation has become a commodity now. Dehlivery can deliver and so can multiple other delivery startups. Businesses can get easy storage. Agents(chatgpt) can create customised checkout experience for buyers on the go, with integration with Stripe. Discovery is easy with Gemini, Gpt, perplexity etc, as a user I never have to visit Amazon again. As there is not much Amazon would add value to a user. ONDC if can create a meta-agent for businesses, Clients like GPt and other can simply place order and checkout in seconds without any API or marketplace. The missing part is trust for Returns, Quality, but that was still always the case before eCommerce, if small businesses can create solid brands from customer experience perspective, they don't need to spend 15% commission to marketplace. Local commerce and quick commerce can bridge the gap, having phyiscal presence at convenient distance brings trust. https://lnkd.in/gnUXPjzw

10

Harsh but the truth. Domain knowledge may help you win in the short term, true product skills help you win in the long-term.

6