whitepaper.id
whitepaper.id works with SAP integration and interface management, helping teams make complex integration landscapes more transparent, manageable and controlled.
As user interface designer, I worked on the design and development of an SAP integration cockpit, using Svelte and IBM Carbon Design System.
01 Understanding the problem
Companies run many systems. Interfaces connect them, and those connections carry the processes that run across applications. This section explains the problem the interface was designed for.
01
A simple network
A few business systems exchange information through interfaces.
02
A complex network
The problem is not connecting two systems. The problem is understanding hundreds of connections at once.
03
A structured, managed network
The same systems and interfaces, grouped, indexed and classified.
- InventoryWhat exists?
- DocumentHow is it connected?
- MonitorWhat is happening?
- GovernHow is it controlled?
- Transparency
- Control
- Quality
whitepaper.id
SAP interfaces · Interface management · Integration excellence
whitepaper.id works with SAP integration and interface management, helping teams make complex integration landscapes more transparent, manageable and controlled.
Integration happens between systems.
Understanding happens through the interface.
02 The integration cockpit
A control center for understanding interfaces, integrations and the systems around them.
If the integration landscape is the system, the Cockpit is the interface through which that system becomes understandable. Complex products do not need more visual invention. They need stronger rules: IBM Carbon provides consistency, hierarchy, predictability and a pattern for every new screen.
Step 01 / 12 · Sign in
Outside the system: a restrained sign-in.
WHINT
Integration Cockpit
Portfolio reconstruction · demo environment
- Sign in: Outside the system: a restrained sign-in.
- Overview: Inside the Cockpit. An overview of the whole landscape, before any detail.
- Catalog: The catalog: applications, partners, objects and interfaces as one structured inventory.
- Integrations: Every integration, with its sender, receiver and transferred object.
- Search: Search narrows the inventory.
- Integration: One integration. The row opens into its detail: sender, receiver, object.
- Components: An integration is not necessarily one line. It can consist of several technical components.
- End to end: The same integration as a path: sender, API proxy, queue, integration flow, receiver.
- Component: Select a component to see what it does.
- Landscape: Zoom out: the integration in its landscape of applications, integration layer, partners and objects.
- Reporting: Reporting: the runtime signal of the same integration.
- Errors: Errors and stability, for exploring the interface. The numbers are demo data.
03 Seeing the integration
One order. Many systems.
What looks like one purchase to a customer can trigger a chain of systems, interfaces and decisions across an enterprise.
Packet ORDER #10482Demo data
- Sender
- Customer · web client
- Receiver
- Commerce · storefront
- Object
- Order
- Event
- CheckoutSubmitted
- Type
- REST API
- Environment
- PROD
- Status
- Healthy
- Step 01, Customer → Commerce: A customer buys a product. The token reads ORDER #10482. Debug reading: Customer · web client sends Order (CheckoutSubmitted, REST API) to Commerce · storefront, PROD, Healthy.
- Step 02, Commerce → Payment: The payment is requested. The token reads PAYMENT #10482. Debug reading: Commerce · storefront sends Payment (PaymentAuthorizationRequested, REST API) to Payment service · PSP, PROD, Healthy.
- Step 03, Payment → Commerce: The payment is processed. The token reads PAYMENT #10482. Debug reading: Payment service · PSP sends Payment (PaymentAuthorized, REST API) to Commerce · storefront, PROD, Healthy.
- Step 04, Commerce → ERP: The order enters the business system. The token reads SALES ORDER #10482. Debug reading: Commerce · storefront sends SalesOrder (OrderCreated, Integration Flow) to ERP Central, PROD, Healthy.
- Step 05, ERP → Warehouse: The warehouse receives work. The item is picked. The token reads PICK REQUEST #10482. Debug reading: ERP Central sends PickRequest (PickRequested, Integration Flow) to Warehouse · WMS, PROD, Healthy.
- Step 06, Warehouse → Loading dock → Logistics: Shipping starts. The token reads SHIPMENT #10482. Debug reading: Warehouse · WMS sends Shipment (ShipmentBooked, REST API) to Carrier API, PROD, Healthy.
- Step 07, Logistics → Notification → Customer: The customer is notified. The token reads NOTIFICATION #10482. Debug reading: Carrier API sends Shipment (ShipmentDispatched, Event) to Notification service, PROD, Healthy. Debug reading: Notification service sends Notification (CustomerNotified, Notification) to Customer · web client, PROD, Healthy.
- Step 08, The complete flow: The business sees one purchase. The integration landscape sees a chain of events. Debug reading: Notification service sends Notification (CustomerNotified, Notification) to Customer · web client, PROD, Healthy.
Two readings
What looks like one purchase to a customer can trigger a chain of systems, interfaces and decisions across an enterprise.
The same process can be read twice. Story shows what the business experiences. Debug shows what the integration landscape is doing underneath. The world does not change; only the reading does.
04 From one to thousands
One process can be followed by eye. A large integration landscape cannot.
Order #10482
One business process. It can be followed by eye.
- 01 Customer → Commerce
- 02 Commerce → Payment
- 03 Payment → Commerce
- 04 Commerce → ERP
- 05 ERP → Warehouse
- 06 Warehouse → Loading dock → Logistics
- 07 Logistics → Notification → Customer
Order #10482: one business process across 7 steps.
Representation
Visualization can narrow the landscape from thousands of relationships to one explainable process.
- One process
- show the world.
- One hundred flows
- show the topology.
- One thousand integrations
- show the structure.
Then filter until the detail becomes human again. The same data supports every view; only the representation changes with the question. No AI is involved: filters, structure and semantic zoom do the narrowing.
Filtering works when you know what to look for.
What happens when you do not?
Reflection
Working at whitepaper.id introduced me to a class of software where the interface has to make highly technical systems understandable without oversimplifying them. That problem has stayed interesting to me: how information architecture, visualization and motion can help people reason about infrastructure they cannot see directly.