Data Lake

The Data Lake is where a workspace keeps the data its automations rely on but do not produce: clients and their agreed terms, product codes, ports, carriers. Keeping it in one place means every automation checks against the same values, and changing a value changes it for all of them.

Find it under Data Lake in the workspace rail. Each workspace has its own tables. Opening and changing them needs the Organization admin or Operations director role.

Reference tables

The Data Lake holds reference tables. Each has named columns with a type, and one primary key column that identifies each row, such as a client code.

Open a table to see two tabs. Data shows the first 100 rows. Columns shows each column’s type, whether it is required and which one is the primary key.

A Clients reference table's Data tab, with rows of client codes, names, Incoterms, currencies and ports of entry.

The same table's Columns tab, listing each column's type and primary key, with the form that adds a nullable column.

Add nullable column on the Columns tab adds a column that may be empty, and existing rows keep working.

Rows come from automations

You never type rows into a table. An automation writes them with Write Reference Table, so every change is repeatable and recorded in a run. A row with a primary key that already exists replaces that row, and a new key adds one.

The rows can come from a CSV file someone uploads, or be entered in the block itself.

The Last load attempt on a table’s Data tab shows how the most recent write went: whether it succeeded, when, how many rows were read and written, and which automation did it. Start here when a table looks out of date.

A product library reference table's Data tab, with a successful last load attempt and rows of SKUs, invoice wordings and commodity codes.

Automations look records up

Two blocks read a table while a run works:

Find Required Reference Record gets the one row whose column equals a value, such as the client named on an invoice.
Read Reference Table gets every row.

A Find Required Reference Record block's settings, matching the consignee from a document against the Client name column of the Clients table.

When a required record is not found, the run can take the block’s Not found exit. Without that exit it pauses and asks a person for the right value, then tries again. See Escalations and approvals.

Archive a table

Archive table makes a table read only and keeps its rows and columns. Automations that use an archived table fail at that block until you click Restore table or point them at another table.

Data Packs

Browse Data Packs lists ready-made table definitions you can install. A pack adds tables and their columns, never rows: you preview the schema changes, then apply them, and nothing runs. You still load the rows with an automation. The list is empty when no pack is available for your workspace.

Look up reference data walks through creating, filling and using a table.
Put a person in the loop