/Cocoa/
A campaign chatbot that turned a selfie into a personalised Pantone colour card — live on WhatsApp, powered by AWS and AI image analysis.
Nivea needed something people would actually use and share. I built a WhatsApp chatbot for their Cocoa Shades campaign that took a user's selfie, ran it through AWS Rekognition to read their exact skin tone, then generated a custom Pantone-style colour swatch matching their complexion — ready to download and post. The whole thing ran on PHP, AWS, and Infobip's messaging platform.
/Project Goals/
Give every user a personalised skin tone moment they'd want to share — no app download, no friction, just WhatsApp and a selfie.
The campaign needed mass reach without a native app. WhatsApp was the right channel — zero install barrier, already on every phone. The bot had to feel natural to talk to, handle selfie uploads cleanly, process images fast, and return something visually sharp enough that people would actually share it. Every step had to work without manual help.
/Results/
Users got a personalised Pantone swatch in seconds — organic sharing did the rest, exactly what the campaign was designed for.
The bot handled the full user journey — greeting, selfie collection, skin tone analysis via AWS Rekognition, dynamic swatch generation via IMGix, and delivery back through Infobip on WhatsApp. Users received a branded, shareable colour card matched to their complexion. The campaign spread organically through user shares, hitting Nivea's goal of personalised reach without paid distribution.
AWS RekognitionAmazon EC2AWS S3PHPIMGixInfobipWhatsApp APIAI Image Analysis