AI content generation for a multi-brand media publisher
Writing product reviews by hand at scale was uneconomical. We engineered prompts for structure, correct Amazon affiliate links and tone, plus an admin panel so editors tune voice without developer involvement.
Client context
A US media group monetises deal and product-review content across many brands (e.g. Amazon Deals-style verticals). They needed ~100 short, uniform reviews weekly without a large editorial payroll.
Challenge
ChatGPT drifted from required structure; product links were often wrong; tone was overly promotional. Editors needed safe experimentation without breaking publish rules.
What we did
- Tuned prompts for headline, two-sentence description and 'why buy' format with strict response schema.
- Solved affiliate link accuracy with explicit catalog instructions and automated structure validation.
- Built React/Node admin panel for tone-of-voice intro experiments; devs own structure and QA gates.
- Integrated Scala services where existing stack required it.
Process
- Prompt iteration with product team on real deal inventory.
- Automated reject for low-quality or malformed drafts.
- Tone panel for non-developer experimentation.
- Scale to ~100 articles/week in production.
Result and impact
Automated review pipeline generating thousands of dollars weekly in ad/referral revenue at existing traffic levels; editorial capacity freed for complex pieces.
This is an NDA-protected engagement. Client name and identifying details are withheld; industry, region, challenge, solution and outcome are shared in an approved form.
We monetise scale without hiring an army of copywriters.Product lead (quote representative, under NDA)