Zepto

Quarterly Report filed 2026-05-14

Material Event filed 2026-05-01

Proxy Statement filed 2026-04-30

At Zepto, we move fast — shipping hundreds of commits across multiple services every day — and our sensitive surface area is extensive: code repositories, documentation platforms like Confluence, issue trackers like Jira, countless messages on communication channels such as Slack, and even serverless functions running critical logic. With this scale, even a single leaked API key or credential can quickly escalate into a security incident or compliance issue. Simply telling people “don’t commit secrets” or relying on a single CI check was not enough. As our development moves rapidly and our tech stack evolves constantly, we needed secret detection everywhere, keeping pace with both our code and infrastructure changes. This led us to build an in-house, end-to-end secret detection system. Here is how we did it: the problem we were up against, the layers we added, how the data flow fits together, and what actually worked. The problem: scale and surface area Secret leaks don’t occur only in

At Zepto, quick commerce means everything moves fast — especially inside our Micro Warehouses. Orders are constantly flowing in, and operations run at high speed.In this environment, there’s no shortage of data. But raw data alone isn’t useful. The real challenge is making fast, reliable decisions. We focused on building a web platform where UI/UX and configurability do the heavy lifting. Instead of static dashboards, teams get clear, real-time views tailored to their role. What matters shows up instantly , no digging, no delays. This makes it easier to catch issues early, respond faster, and maintain quality even at scale. In quick commerce, speed isn’t just about deliveries, it’s about how fast you can understand and act. Operations teams don’t just need dashboards. They need: Consistency : Is this view current? Context : What caused this anomaly? Determinism : If I click this, what exactly happens? Low cognitive overhead : No context-switching across multiple tools To solve this, we

What Is Product Enrichment? When a client requests a product on Zepto, the response isn’t a single database read. It’s an orchestration across a dozen-plus downstream services — catalog, pricing, inventory, ratings, tags, variants, attributes — all assembled in real time, shaped to exactly what that client needs. A full product card brings all of it together: the name and image from the catalog, the discounted price from pricing, live availability from inventory, ratings and review counts, tags like Buy Again and Bestseller , flavor attributes, variant groupings. But not every caller needs all of that. A post-order confirmation screen just needs a name and an image. An internal service might only need inventory. That layer — assembling the right product data for the right caller at the right time — is what we call product enrichment. At Zepto’s scale, product enrichment serves 1.5 million batch reads per minute , powering product listings, cart pages, search, and recommendations. Getti

Model Context Protocol (MCP): The New Way of Shopping AI assistants are everywhere now. We use them to write code, draft emails, plan trips. But what if they could also do your grocery shopping? Not just recommend products, but actually search, add to cart, and place an order on your behalf. MCP (Model Context Protocol) makes this possible. It provides a standard way for AI clients like ChatGPTWhat if your AI assistant could do your grocery shopping? Not just recommend products, but actually search, add to cart, and place an order. We built exactly that. Zepto’s MCP server turns ChatGPT, Claude, and other AI clients into a full shopping experience. Getting here meant solving some new fun problems: a custom OAuth provider for MCP, embedded visual widgets that render your cart right inside the chat, personalised search that remembers your favourite brands, and enabling payments for the true conversational experience. What started as a simple experiment quickly grew into something bigger,

Customer Delight at Zepto In October 2024, Zepto set out to build a customer support platform that could deliver hyper-personalized experiences at scale. What emerged was a robust, multi-tenant support platform powered by AI-driven personalisation — achieving 75% ticket automation while maintaining 90%+ customer satisfaction, now handling consumer support, delivery partner support, and merchant support across 10x traffic bursts during critical business moments. This is the story of how Zepto transitioned from a third-party vendor to owning the entire in-house stack, and the technical decisions that made AI-first, personalised support possible. The situation was clear: Zepto’s support infrastructure wasn’t keeping pace with its growth. Old Architecture The pain points were real: Velocity Crisis : The chatbot was built and maintained by the vendor. Every iteration meant raising tickets, waiting for their engineering team, and watching days turn into weeks. Scale Mismatch : As Zepto scale