Woolworths has launched an artificial intelligence-powered food assistant designed to help South African customers decide what to cook and buy.

Called My Woolies Chef, the new platform is integrated into the Woolworths app. It uses conversational AI to recommend meals, assist with planning and help customers shop for ingredients.

The assistant draws from 20 years of Woolworths TASTE recipes. Customers can request ideas based on the food they already have, dietary preferences, household needs or upcoming occasions.

For example, users could ask for a quick family dinner, a vegetarian meal for two or lunchbox ideas that can be prepared in advance.

Limited Testing Starts in September

A select group of existing MyDifference loyalty programme members will receive access from September 2026.

Woolworths will use this beta testing period to study customer behaviour, improve recommendations and refine the overall experience. A broader rollout is planned for early 2027.

Woolworths Group Data and AI Officer Jose Rodrigues said the platform allows customers to explain what they need using everyday language instead of completing several separate searches.

Recommendations will consider locally available ingredients, seasonal relevance and practical cooking needs for South African households.

Recipes Connected to Online Shopping

My Woolies Chef will also connect recipe planning directly to Woolworths’ online shopping service.

Once customers choose a recipe, they can add the required ingredients to a linked Woolies DASH basket. They can then remove items already available in their fridge or pantry.

Woolworths said this could help shoppers use what they already have at home and avoid buying unnecessary ingredients.

Rodrigues said the company wants to use AI to solve practical customer problems and improve how people plan, shop and cook.

The retailer believes the platform could eventually support more personalised meal plans, smarter shopping lists and recommendations based on customers’ budgets and needs. Future features will partly depend on feedback received during the testing phase.