A WhatsApp-native order and campaign management platform for SME vendors in emerging markets. Embary took it from concept to live product, handling product strategy, architecture, Twilio infrastructure, and full build.
MessageBench came to Embary as an observation: tens of thousands of small vendors across West Africa were running their entire businesses through WhatsApp, with nothing but chat threads to track orders, chase payments, and manage customers. The opportunity was clear. The product didn't yet exist.
Embary was engaged to define what the product needed to be, design the system architecture around the WhatsApp Business API, integrate Twilio as the messaging backbone, and build the full application: the Workbench order tracker, the Studio broadcast engine, and the automation layer that connects them.
The brief was to build something vendors could pick up in an afternoon and see results by end of week. That constraint shaped every decision from the data model to the interface.
WhatsApp commerce in emerging markets isn't informal by accident. It's informal because no one built the infrastructure layer. Vendors fielding 40 to 60 orders a week were managing everything through conversation threads, voice notes, and payment screenshots buried in DMs.
The technical challenge wasn't novel. The product challenge was harder: build something that slots into an existing workflow without asking vendors to change how they operate. Any friction and the product would fail, regardless of how good the underlying system was.
Rather than pulling vendors out of WhatsApp, MessageBench sits alongside it. Incoming messages surface in the Workbench as structured order records. Payment confirmations trigger status updates. Follow-ups fire automatically. The vendor's customers never know anything changed.
Twilio handles the messaging infrastructure. The architecture was designed to be stateless at the API layer and durable at the data layer, so a vendor can run 200 active conversations without dropping a single order to a missed notification or a slow query.
User research with active WhatsApp vendors to map the real workflow, not the assumed one. Output: a product spec with clear scope boundaries, a prioritised feature list, and the core constraint that everything must feel native to the WhatsApp context.
System design for the full stack: Twilio webhook integration, order state machine, conversation routing logic, broadcast segmentation engine, and the data model that ties them together. Designed for multi-tenancy from day one.
Full product build: the Workbench order tracker with real-time status management, the Studio broadcast engine with audience segmentation and delivery tracking, and the automation layer for payment reminders and shipping updates.
Onboarding of founding customers, monitoring infrastructure performance under real usage, and iterating on the interface based on vendor behaviour in the first 60 days. The product launched to paying customers and has been in continuous operation since.
MessageBench launched on schedule and is operating with paying customers across West Africa. No post-launch architecture changes required.
Early customers moved from losing 40 or more orders per month to zero within 60 days of onboarding. The order state machine holds.
Founding customers report significant revenue growth in the first 90 days, driven by broadcast campaigns and the elimination of order drop-off.
I had customers sending me payment proofs on a Tuesday and I'd find them on Friday buried under 200 other messages. MessageBench changed that in a week. Those orders don't disappear anymore.
FashionVault — Founding Customer · Lagos, Nigeria
A multi-jurisdictional digital asset platform for green energy credits, delivered end-to-end across business model validation, regulatory mapping, and full product build.