Files · CSV
Integrate CSV files. Turn exports into an orderly data flow.
Price lists, product records and stock often arrive as CSV files. With onflAI you describe the intended transfer and make column mapping, checks and processing reviewable.
A CONCRETE EXAMPLE
What your requirement could look like.
“Import our supplier's daily CSV price list. Update products by SKU and flag incomplete rows.”
Current prices in the target system instead of repeated manual spreadsheet imports.
Planning example, not a customer case study. Execution, scheduling and scope depend on connectivity, approval and plan.
01
Prepare access
- Confirm the file location, naming pattern and available transfer method. A file format does not define its transport.
- Provide an anonymised sample including the header. Agree character encoding, delimiters and decimal notation.
- Define how a fully delivered file is identified. A partially written file must not be treated as a complete inventory.
02
Define fields and rules
- Map columns such as SKU, price and currency to target fields. Preserve leading zeros in product identifiers.
- Account for quoting, embedded delimiters and line breaks. Empty fields do not automatically mean the numeric value zero.
03
Check before production
- Test malformed rows, duplicate SKUs and unexpected columns. Decide whether an error rejects the entire file or only an individual record.
- Prevent an already processed file from being booked unintentionally again. Plan file identification, processing results and restart behaviour together.
How are different CSV formats handled?
The file profile defines delimiters, character encoding, headers and date formats. Representative test data helps ensure values are interpreted correctly and transferred into the right target fields.
Where onflAI helps
You describe the outcome and select the relevant data. AI assists with configuration, mapping and validation. Credentials remain separate and are processed in encrypted form by the platform. After review and human approval, IOnbrix executes the data process, not AI.
A validated configuration does not replace an end-to-end test with your systems. Verify network access, the correct business outcome and error behaviour before production.