ERP & SAP

Typical systems

• SAP
• other ERP and business backbone systems

Realistic start strategy

1. do not start with full-custom integration
2. start with scheduled exports
3. validate first use cases via Data Pools

Looking up records: limits of a data source

A data source searches documents, not records. Tables (Excel and CSV) are not indexed row by row: at most 2,000 rows per sheet, 5,000 per file and 64 columns are searchable. Search does not find rows beyond that, so an export with more records (for example all offers) cannot be queried completely by offer number or customer name. Details: Formats & Access.

For targeted lookups by number or name:

• Live query: the Database tool or a Custom MCP Server directly against the ERP or its database
• Analyzing large tables: load the export file into the Code Sandbox
• Documents as a source: for a manageable number of offers, one document per offer (number in the file name and in the text) instead of one collective table, see Preparing Your Data. As a rule of thumb, connect 500 to 1,000 relevant files per data source.

When deeper integration is required

• add targeted live integration via Custom MCP
• only go deeper where business value is proven

Practical tip

Ship 1-2 stable use cases first instead of waiting months for a perfect full-stack integration.