Unattended running is the real prize
The portal cannot run overnight. The API can. Automation is why organisations move to the API in the first place — but unattended running only works if the exception path is designed as carefully as the success path.
Start with what should never be automated
Some categories of movement always need human review — new suppliers, unfamiliar commodity codes, high-value consignments, licensed goods. Define these before automation, not after. The automated TSS submissions guide makes the same point about hybrid coverage.
Response handling
Responses come back as JSON. Your system needs to interpret success, partial success and failure states, log them, and act. Silence is not success.
Reconcile against source data
Every automated submission should be reconciled against the source record that triggered it. Missed submissions and duplicated submissions are the two biggest hidden risks.
Guardrails
Rate limits, retry caps and duplicate detection stop a bad batch turning into a bad month. Design them in, do not bolt them on.
Observability
Dashboards for submission volume, rejection rate and turnaround time. The team should know within minutes when something changes.
See TSS automation with guardrails and hybrid fallback in one demo. Watch a demo.
Top comments (0)