Hinglish AI Seller Operations Platform for Indian Marketplaces

A Hinglish voice-and-chat AI assistant fused to a complete seller-operations back office for Indian marketplaces - Amazon, Flipkart, Meesho, Myntra and Shopify. Ops staff ask for work in the language they actually speak, and it gets done.

The differentiator is the write path. Indian marketplaces give third parties no dependable API for changing stock and prices, so this does it another way: every write enters an approval queue with a preview, and a Playwright executor then drives the seller's own logged-in portal session. No marketplace API keys are needed at all, and nothing reaches a marketplace until a human approves it. For an agency running clients' accounts, it also means never holding a client's credentials.

Around that sits the rest of a seller's day: GST invoicing with GSTR-1/3B and e-way bill payloads, TDS reconciliation and Tally export, SAFE-T claims and courier weight-dispute recovery, NDR and RTO handling, repricing against a margin floor, rank tracking, ad bid optimisation, suspension-risk scoring and appeal drafting, B2B quotations, SKU-level P&L, multi-tenancy with RBAC, and an Android app. 619 route handlers across 42 router modules, in 18 functional areas.

Reproducible from the repo: 5,057 backend tests passing in CI, 78% line coverage over a build-failing 70% floor, a per-module coverage ratchet and an architecture import-direction ratchet, 100 regression tests each named for a real defect, 130,000 lines of Python and 24,000 of JS/TS.

It is pre-revenue. Zero customers, zero revenue, no production hours - built and tested, not yet running for a paying seller. That is the most important fact about it, said here rather than left for diligence.

Sold outright with full IP assignment. One author, no contractors, no third-party contributors.

The product tour, capability schedule and a 90-second recording are public and live - screenshots and video in the gallery, full links in the seller's notes.