.png)
Research report
AI Does Not Remove the Bottleneck
Find your real workflow constraints before buying AI tools.
Composable solved the monolith. It created an integration problem.
Commerce platforms, order management and the operational systems that carry a peak trading day.
Where we help
The work we are asked to do in retail & e-commerce.
01
Composable and headless builds
MACH architecture assembled so the seams are owned rather than discovered.
02
Order management
The reconciliation layer that quietly decides whether the customer gets their parcel.
03
Peak readiness
Load modelling and cost control for the day the year is decided on.
04
Unified customer data
One view across online and store that the merchandising team can query.
05
Store and fulfilment systems
Tools for the people picking, packing and serving in store.
The numbers behind our retail & e-commerce work
Composable commerce solved the monolith problem and created an integration problem. Every MACH stack we have inherited has the same weak point: order management, doing quiet reconciliation work that nobody scoped and nobody owns. It surfaces as a support queue rather than an architecture decision, which is why it stays unfixed. We scope it explicitly, before the first vendor is chosen.
1.2M sessions
the load we have held
+21%
what it moved
15
years building for retailers
Selected work
Engagements in retail & e-commerce, described rather than named.
A US medical equipment resellerBuilding a sales channel the business could ownA storefront of their own beside a high-volume eBay business, with inventory kept in step both ways and double sales resolved automatically.5 minfor stock to reflect across both sales channelsRead the case
A safety training providerTurning every sale into an enrolmentShopify and direct checkout feeding a single SCORM platform that provisions accounts, tracks progress and issues certificates unattended.Zeromanual admin steps between a sale and a live enrolmentRead the case
A Los Angeles private car serviceTurning every call into a booking.Phone and web bookings answered by AI, written straight into the dispatch system, then rebuilt as a product other operators can buy.24/7booking cover on phone and chat, with nobody on the lineRead the caseQuestions we get asked
Can you handle peak trading?
Peak readiness is a programme that starts months before the day. Load testing against realistic baskets, dependency mapping so a slow third party degrades instead of cascading, and a freeze plan everyone agreed in advance. The failures we are called in after are almost never the storefront itself.
We are on a composable stack. Can you work with it?
Yes, and we usually find the weak point in the same place. Composable solved the monolith and created an integration problem, and order management is where it concentrates — quiet reconciliation work that nobody owns until something goes wrong on a Saturday. That is where we tend to start.
Should we replatform or improve what we have?
Improve, unless the current platform is genuinely preventing something the business needs. Replatforming consumes a year and usually buys back the position you already held. We look for what is actually blocked — a channel you cannot launch, a market you cannot enter — before proposing a rebuild.
Can you connect our online and in-store systems?
That is most of what retail engineering turns out to be. Inventory, orders, returns and customer identity each exist in more than one system with different rules, and the work is deciding which one is authoritative for each. We settle that before building anything on top.
How do you avoid overselling stock?
By treating inventory as a distributed-systems problem, which is what it is. Near-real-time synchronisation between channels, durable event processing so nothing is lost on a restart, a reconciliation sweep to catch drift, and automatic handling for the sale that slips through anyway.
How quickly can you move?
10 wks to a first release, and the constraint is usually integration access rather than build time. Getting credentials and test environments from an existing platform vendor is the step that most often sets the timeline.
How we deliver it
.png)
.png)
