How your inputs are used

The create screen asks for four inputs. They are not passed to the generator as written — they are turned into specific instructions for each of the seven fields.

Diagram: purpose, brand voice, search keywords, and additional context are applied to every field, producing per-field instructions — for example a description written warm and benefit-led, an SEO meta description that is keyword-aware and within length, and conversational Q and A giving direct citable answers.
One set of inputs produces a separate instruction set for every field.

Why this matters for how you write them

A description and an SEO meta description have different jobs, lengths, and readers. Your inputs have to serve both. “Friendly” gives every field almost nothing; “warm and plain-spoken, never uses exclamation points, says sofa not couch” gives each field something concrete to apply.

The practical consequence: vague input does not produce slightly worse copy in one place. It weakens all seven field instructions at once, which is why a first batch that reads flat is usually an input problem rather than a product problem.

The four inputs

InputWhat it controlsCovered in
PurposeWhich outcome each field is written towardPurpose, keywords, and context
Brand voiceHow the content soundsWriting brand guidelines
Search keywordsTerms worked into copy where they fitPurpose, keywords, and context
Additional contextAnything specific to this batchPurpose, keywords, and context

Reuse and iteration

You can repeat an enrichment on the same products with new brand voice and context. That makes inputs cheap to iterate on a small batch and expensive to get wrong on a large one — tune on 25 products before committing 250.