Working with large data
Many Wrkflows start with a single record and end with one result. Real automation often works the other way around: hundreds of spreadsheet rows, paginated API responses, or a folder of attachments that all need the same treatment.
Wrk gives you two complementary ways to scale that work:
- Divides — built into certain Wrk Actions so the platform can process many independent items in parallel
- Looping — Wrk Actions you wire together to step through a list or numeric range one item at a time
When to use which
| Divides | Loops | |
|---|---|---|
| Who controls it | Mostly automatic — the platform splits work inside a dividing Wrk Action | You design it — pair a loop Wrk Action with Continue Loop |
| Best for | Large batches of similar, independent tasks (enrich each contact, download each attachment) | Ordered or conditional iteration (paginate an API, process items until a condition is met) |
| What you see in the Wrkflow Designer | A single Wrk Action card; divide behavior is indicated on that Action | A loop branch that connects back to the loop Wrk Action |
| Throughput | High — items run concurrently | Sequential by default; nested or concurrent patterns are possible with care |
You can use both in the same Wrkflow. For example, a loop can walk through API pages while a dividing Wrk Action inside the loop enriches every row on that page at once.
Go deeper
- Divides — how to spot dividing Wrk Actions and when parallel execution helps
- Looping — the loop + continue pattern and what it looks like in the Wrkflow Designer
- Loop basics — hands-on patterns for pagination and JSON arrays
- Speed up your Wrkflow with Divide — extended how-to on divide performance
- Launch execution — how branching, parallel paths, and divides fit into Launch order