AQNEX E-Commerce Suite

Three systems that did not talk

AQNEX was running a retail operation across three disconnected pieces of software. Customers shopped on a legacy storefront. Staff worked orders in a separate back-office dashboard. Suppliers sent inventory as spreadsheets over email, which someone then re-keyed by hand.

Each system was individually defensible and collectively unworkable. Stock counts on the storefront reflected whenever the last CSV had been imported, so customers bought items that were already gone. Dispatch could not see supplier lead times. Reconciling what had actually been sold against what suppliers thought they had shipped was a weekly manual exercise.

The brief was not "make these three talk to each other." Integrating them would have preserved three sources of truth and added a fourth problem. The brief was one platform, one data model, three ways in.

One write path

The suite is a single application with three portals over shared data.

The storefront is what customers see: search, dynamic filters, product pages and a checkout flow designed by removing steps rather than adding reassurance to them. It is mobile-first, because that is where the traffic was.

The admin hub is where the business runs. Live sales monitoring, the dispatch queue, catalogue management, and role-based permissions so warehouse staff and finance see different things.

The supplier portal is the piece that did not exist before. Verified vendors manage their own listings, batch-upload inventory, and track fulfilment against their own orders. The spreadsheets stopped, along with the transcription errors that came with them.

Because all three write to the same records rather than syncing between copies, stock cannot drift. What a customer sees, what dispatch sees, and what the supplier holds are the same number by construction — there is no reconciliation step to skip.

The stack, and why

Next.js and Node with PostgreSQL underneath, GraphQL between them. The interesting choice was PostgreSQL over a document store: inventory is inherently relational and inherently concurrent, and the guarantee we needed most was that two people cannot sell the last unit at once. Transactional integrity was the requirement the rest of the architecture arranged itself around.

GraphQL earned its place for a specific reason: three portals with very different data appetites reading from one schema. The storefront wants a shallow product view fast; the admin hub wants deep, cross-referenced operational data; the supplier portal wants a narrow slice scoped to one vendor. One endpoint, three query shapes, no proliferation of bespoke REST routes.

Demonstrating it without exposing it

One requirement was unusual enough to be worth naming. AQNEX needed to show the platform — to partners, to prospective suppliers, at trade events — without exposing real revenue figures and real customer records to a room.

So the platform has a presentation mode: a toggle that masks transaction values, customer identities and sales aggregates while leaving every workflow fully functional. It is a small feature that removed a real constraint on how the business could talk about its own tools.

Where it stands

The suite is live and running the operation. Manual CSV exchange is gone, inventory is synchronised in real time across all three portals, and the three systems that did not talk are now one system that does not need to.