Ask Sim to create a workspace table, inspect its data, or change rows. Reference the table by name or with an @-mention. Tables can open in Chat's resource panel for review.
Table reads, writes, creation, and exports follow your workspace access and organization policy. Storage and row limits depend on the plan and deployment configuration.
Describe the columns and types, or attach a CSV for Sim to import:
- "Create a leads table with name, email, company, status, and created date columns."
- "Create a table matching this CSV." Attach the file.
- "Create an errors table with message and workflow text columns, a timestamp date column, and a resolved boolean."
Inspect the schema and imported rows before using the table in an agent. Specify unique or required columns when the data needs those constraints.
Name the table, fields, filters, and time range:
- "How many leads have status 'qualified'?"
- "Show leads from the past seven days with score above 0.8."
- "Find duplicate emails in the contacts table."
- "Calculate conversion from 'contacted' to 'meeting booked' in this month's pipeline data."
Sim can query the matching rows and compute summaries. It can also read multiple tables and combine their results: describe the relationship, such as matching company_id between contacts and companies. This does not create a persistent join between the tables.
Chat can mask tracked secret values or omit rows and selected cells it cannot safely check. Matching-row totals and pagination cursors still describe the database query; a shorter response does not prove that every match was shown. Summaries based on returned rows can be incomplete.
Large reads may need pagination or an export. Name the subset you need instead of requesting every row when a filter can answer the question.
Provide values by column name. For bulk changes, state the condition:
- "Add a leads row with company=Acme Corp, email=jane@acme.com, and status=pending."
- "Mark queue rows as processed where createdAt is before today."
- "Set price to 49 for rows where tier is 'pro'."
createdAt is the built-in creation timestamp. Use the actual custom column name when you mean a separate date field.
Review the affected records and ask Sim to verify the updated values. Table writes can also trigger configured workflow automation.
A write can be saved even when its returned row is withheld from Chat. Inspect the table before repeating a change whose result could not be shown.
Deleting rows removes their stored data. Identify the table and condition before asking for a destructive action:
- "Delete all rows in the test_events table."
- "Delete contacts with status 'test' from the contacts table."
Removing rows does not undo external actions previously performed by workflows using them.
Export the whole table or a filtered subset as CSV:
- "Export the leads table as leads.csv in workspace Files."
- "Export pipeline rows where status is 'closed' and save them as closed-deals.csv."
Wait for the export to complete, then inspect its row count and download the file. A CSV created from rows returned to Chat can omit rows unavailable to Chat. Use the table's own export control when you need an authorized export of the stored data. A saved workspace CSV can be referenced by other workflows.
Ask Sim to configure a Table block or tool with the intended table and operation, or select the table in the workflow builder. The table must be available to the workflow's execution context; naming it in Chat does not replace the block's configuration.
For per-row enrichment, ask Sim to connect a deployed workflow to table columns so its outputs populate each row. See workflow columns for dependencies and execution behavior.