Data transformation prepares information for processing or storage. In an AI workflow, records from spreadsheets, CRM systems, databases, or other digital systems can be reshaped into inputs an LLM can process, while generated text can be reformatted for downstream use. section 7.0
A practical transformation workflow can:
- Extract data from files, databases, or enterprise applications. section 7.4
- Load, join, sort, aggregate, and reshape records into formats suitable for LLM prompts. section 7.4
- Split generated text into separate fields for database storage or downstream systems. section 7.0
Customer-record processing
A CRM table containing customer names, ages, and recent purchases can feed a reusable prompt across many records. The resulting text can then be split into fields and stored for later use. section 7.0
Figure 7.1 The LLM’s input text might come from various sources—files, systems, or users.
Shows how LLM input text can originate from files, systems, or users. section 7.0
