Product · WhatsApp commerce — In development
Bapoojee
Turn WhatsApp chats into paid orders — one shared inbox for your whole team.
Bapoojee is a WhatsApp commerce platform built for East African shops that already sell through chat. It gives a team one shared inbox, a product catalogue, order tracking and payment collection, so a conversation can go from “How much?” to “Paid” without ever leaving WhatsApp. AI works quietly in the background, sorting incoming messages, routing them to the right person and drafting replies.
Industry
Retail & commerce (WhatsApp Business / SaaS)
Role
Product strategy · Design · Engineering
Status
In development
The challenge
Across the region, an enormous amount of retail already happens on WhatsApp — but it happens on one person’s phone. There’s no way for a team to share the workload, no record of who promised what, and no link between a chat and an actual order or payment. As a shop grows, messages get missed, orders get lost in scrollback, and the owner becomes the bottleneck. Off-the-shelf helpdesk tools assume email and a Western checkout flow; they don’t match how these businesses actually sell.
What we built
We built Bapoojee as a multi-tenant SaaS product: a shared team inbox wired directly into the official WhatsApp Business platform, with a catalogue, orders, payments, broadcasts and a customer database behind it. Every workspace is isolated, every action is logged, and the whole system is built to be operated by a team rather than an individual. An AI layer handles triage, routing, suggested replies and a self-serve knowledge base — metered per workspace so AI costs never run away.
Key features
One shared inbox
The whole team answers from a single WhatsApp number, with assignment, notes, labels, snoozing and read tracking, so nothing gets missed or answered twice.
Chat-to-checkout
Send a product card, take the order and send a payment link inside the same conversation; the order updates itself when payment lands.
Product catalogue
Manage products and categories once and push them straight to WhatsApp, so customers browse and buy without a separate website.
Order tracking
Every order has a status the customer can check by asking; status changes can notify them automatically.
Broadcasts
Send approved WhatsApp campaigns to a chosen audience, with opt-out handled automatically and per-recipient delivery tracking.
Automations
Welcome messages, out-of-hours replies and keyword responses that respect business hours and don’t spam repeat customers.
Customer records
A full customer view with custom fields, segments, import/export, duplicate detection and merging.
AI assistance
Incoming messages are classified and routed to the right agent, common questions are answered from a knowledge base, and agents get suggested replies.
Guided conversation flows
Build step-by-step chat journeys — ask a question, branch on the answer, tag the customer, hand off to a human — without writing code.
Admin & audit
Connect WhatsApp numbers, manage the team and roles, review message templates, and see an append-only log of every privileged action.
Engineering highlights
Built to survive real-world messiness
Incoming WhatsApp events are verified, saved and queued instantly, then processed in the background — so a burst of traffic or a slow third party never drops a customer’s message. Duplicate messages are ignored automatically.
Credit-metered AI with a built-in margin
Every AI call reserves credits before it runs, refunds what it doesn’t use, and records what it actually cost — so the platform can never be run at a loss, and prices can be adjusted without a redeploy.
Respects WhatsApp’s rules by design
WhatsApp only allows free-form replies within 24 hours of a customer’s last message; the system derives that window live from the conversation rather than storing it, so it can never drift out of sync and trigger a policy breach.
Multi-tenant from the ground up
Every record belongs to a workspace, and isolation is enforced at the database layer — one shop can never see another’s customers, even by accident.
Quality gates on every change
Automated formatting, linting, strict type-checking, secret scanning, dependency audits and a 90% test-coverage floor on the API run on every commit.
The stack
- Frontend
- Next.js · React · TypeScript · Tailwind CSS · shadcn/ui
- Backend
- NestJS · PostgreSQL · Drizzle ORM · Zod · Better Auth · Graphile Worker · Redis
- Infrastructure
- Turborepo + pnpm · Docker · Dokploy + Traefik · Hetzner Object Storage · Sentry · PostHog · GitHub Actions CI
- AI & integrations
- Anthropic Claude · OpenAI · WhatsApp Business Platform · Stripe · Pusher · Resend
Next project