Begin with the catalogue: find your kind of system, its usual direction and cadence, and where its credentials live. The next two pages are the rules every integration follows: a retry policy that survives the other system being down, with dead letters that someone is told about, and field ownership, which keeps a two-way sync from looping. The last page puts all of it together: an order pushed to a fulfilment warehouse when it is paid and its status pulled back on a schedule, with idempotency, retries and a daily reconciliation.

The pages build on scheduled jobs, the previous unit, and on the HTTP Request action (HTTP request and response).