Retail Giant Introduces Advanced AI Oversight as Mandy and Olive Evolve into Autonomous Shopping Assistants
Australian retail giant Woolworths is taking another major step in its artificial intelligence journey by expanding the capabilities of its AI-powered customer assistants. The company has announced plans to transform its Mandy assistant into a fully agentic AI system while extending the rollout of Olive, its customer service assistant, to consumers after a successful internal deployment.
The initiative demonstrates how retailers are moving beyond generative AI toward autonomous systems capable of completing customer tasks, making contextual decisions, and managing end-to-end shopping experiences.
Mandy Set to Become a Fully Agentic AI Assistant
Woolworths plans to enhance Mandy, which is already integrated across its Everyday Rewards program, mobile services, and insurance offerings, by equipping it with agentic AI capabilities.
Unlike traditional virtual assistants that respond to simple queries, the upgraded Mandy will be designed to understand customer objectives, execute multi-step tasks, and provide more personalized support across multiple services.
The move reflects the retailer’s broader strategy of embedding intelligent AI agents throughout its digital ecosystem to improve customer engagement and streamline everyday interactions.
Olive Handles Majority of Customer Support Requests
Alongside Mandy’s evolution, Woolworths is preparing to make its AI assistant Olive available to customers after an internal rollout.
According to the company, Olive already manages more than 70% of customer contact center interactions, significantly reducing the workload for human support teams while improving response times.
Initially introduced for employees, the assistant is expected to become available to shoppers, enabling faster assistance with product inquiries, order support, and general customer service.
Eight AI Judges Protect Every Customer Response
One of the most notable aspects of Woolworths’ AI strategy is the development of an internal governance framework designed specifically for autonomous AI.
Rather than relying solely on built-in safeguards from AI models, the retailer has created eight proprietary AI validation agents, referred to internally as “agentic judges.”
Every response generated by Olive is automatically reviewed before it reaches the customer.
Each judge performs a different validation task, including:
✔ Verifying mathematical calculations such as pricing, discounts, and quantities
✔ Checking product information against legal, food safety, and regulatory requirements
✔ Confirming that customer requests have been fully completed
✔ Monitoring response accuracy and consistency
If any validation fails, the system blocks or flags the response before it is delivered, adding an additional layer of quality assurance.
Governance Designed for Enterprise-Scale AI
As organizations deploy increasingly autonomous AI systems, governance has become one of the industry’s biggest challenges.
Woolworths’ multi-agent validation approach demonstrates how enterprises can improve trust in AI-generated decisions without limiting automation.
Instead of treating AI governance as a single security layer, the retailer distributes responsibility across multiple specialized validation agents, allowing each one to verify a different aspect of the response.
This architecture helps reduce errors while increasing confidence in customer-facing AI applications.
Powered by Google’s Enterprise AI Platform
Both Mandy and Olive are built on Google Cloud’s Gemini Enterprise for Customer Experience (GECX) platform.
The platform enables organizations to develop AI agents capable of understanding customer intent, retrieving enterprise information, planning tasks, and interacting with backend business systems.
By combining advanced language models with enterprise integrations, Woolworths aims to deliver more natural conversations while automating increasingly complex customer service workflows.
Preparing for the Future of Agentic Commerce
The retailer has also outlined longer-term plans for AI-driven shopping experiences.
Future capabilities may include proactively preparing customers’ weekly shopping baskets based on purchasing history and allowing shoppers to refine recommendations through conversational interactions.
Because grocery purchases often follow recurring buying patterns, AI agents could eventually anticipate customer needs and simplify routine shopping decisions.
Although these features remain under development, they highlight how agentic AI could reshape digital retail experiences over the coming years.
A Blueprint for Responsible AI Adoption
As enterprises continue investing in autonomous AI, Woolworths’ approach illustrates that successful deployment depends not only on intelligent models but also on strong governance and oversight.
By combining autonomous customer assistants with multiple layers of validation, the retailer is addressing concerns around accuracy, compliance, and customer trust while demonstrating how agentic AI can operate safely at enterprise scale.
The company’s strategy may serve as a model for other retailers and customer-facing businesses seeking to balance innovation with responsible AI deployment as autonomous systems become a core part of digital commerce.
