Case study · Business operations
Connected Business operations simulator
A working model of how requests move between customers, departments and management: reference numbers, routing, approvals, SLA timers, escalations and reporting.
Engagement facts
- Deployment status
- Interactive demonstration. A working demonstration with fictional data, not a deployed product. A working demonstration on this website. It is not a deployed product.
- Client identification
- Fictional organisation. There is no client. The organisation, people and requests are fictional.
- Operational context
- A fictional South African organisation of about 400 people, built for this website.
The business problem
- In a 400-person organisation, requests arrive by email, WhatsApp, phone and in passing. Nobody can say how many are open, who has them, or which are late.
Why it matters
- Late requests surface as complaints instead of alerts. Managers find out last.
The solution
- A request system where every enquiry gets a reference, a department, an owner and a deadline, and nothing changes state without a recorded event.
The architecture
- A pure TypeScript state machine: typed commands in, typed events out, impossible transitions refused with a reason.
- A controlled simulation clock, so SLA breaches and escalations are reproducible.
- Management figures are derived from the event history, never stored separately, so they cannot drift.
Workflow logic
- Intake from any channel creates a reference and classifies the request.
- The request is routed to a department, assigned an owner and given an SLA.
- Approvals go to the right manager; a request that misses its SLA escalates on its own.
- Every change is an event, and the management report is built from those events.
How it was tested
- Automated unit tests on every build for routing, approvals, rejections, SLA escalation, impossible transitions and determinism.
- Browser tests of every view, run before releases as a practice; the build itself enforces only the unit tests.
Verified outcomes
The state machine is covered by automated unit tests for routing, approvals, rejections, SLA escalation, impossible transitions and determinism.
How it was measured: Development. From October 2026. Sample: Desktop and mobile viewports for browser tests. Method: Vitest unit tests run inside the build command; Playwright browser tests run before each release. Tests prove the demos behave as designed. They are not evidence about client systems.
Limitations
- It is a model, not a deployed product. Nothing is saved; a refresh starts over.
- Classification uses fixed keyword rules so it behaves the same every time. A real build would use the client's categories and possibly a classifier with human review.
The demonstration
Interactive demonstration. Fictional organisation and data.
Run the simulatorTell us where the work piles up.
Fifteen minutes on a call or WhatsApp. You describe the job that keeps landing on a person. We tell you honestly whether a system should do it, and what that would take.