/cdx → Azure → multimodal-ai-shopping-assistant
Multimodal AI Shopping Assistant
Part of the scenario: Innovate with Azure AI Apps & Agents - The Customer Journey
This demo highlights how Microsoft AI Foundry as a platform and Foundry IQ as it's knowledge layer is used to embed AI agents in customer-facing Agentic AI solutions—showcasing multimodal capabilities to respond to text, image and video inputs. Runs in a live demo tenant with realistic data — you try the real thing, not a slide deck.
More demos in this story
6 relatedAutomatically Redact PII when Saving Data
This demo showcases how Personally Identifiable Information (PII) data can be protected from unauthorized access by leveraging a feature in Microsoft Fabric to automatically redact any such data (like date of birth, SSN, Account Number) which can be used to identify an individual, before saving it.
Multi-source Retrieval for Agents using Foundry IQ
This demo is a compelling use case to show a core capability in Microsoft Foundry IQ called Multi-Source retrieval. It allows AI agents to pull, combine, and reason over data from multiple different sources at the same time, instead of just querying one database or index. The business benefits include faster and more accurate responses.
Publish Agents to Microsoft 365 Copilot and Microsoft Teams
Frontier organizations need the ability to make the AI agents defined in Microsoft Foundry to be available to data citizens in M365 Copilot and Microsoft Teams. Here in this demo, we see how Zava employees are enabled to freely converse with custom AI agents in M365 Copilot to get proactive shift assignment issues, related business impacts, make shift assignment changes and save instructions to policy changes for handling similar future shift related issues. All this is powered by Work IQ which leverages the AI agents published to M365 Copilot.
Responsible AI and Enterprise-Grade Security
This demo highlights how Microsoft AI Foundry as a platform and Foundry IQ as it's knowledge layer is used to embed Responsible AI practices to protect customer-facing Agentic AI solutions—showing effective defense against DAN-style attacks through guardrails, monitoring, policy enforcement, and continuous evaluation to ensure ethical, fair, and enterprise-ready AI.
Seamless Agent Orchestration Across Channels
This demo shows the power of Copilot Studio to really democratize the use of AI agents by data citizens across the enterprize regardless of their channel. This really puts the ability of defining and leveraging personalized AI agents across channels in the hands of both technical as well as business users too!
Session Management for Preserving Conversation History
Customers frequently make their purchase decisions over multiple chat sessions. This demo shows how the conversation history of a previous session chat between an AI agent and customer can be leveraged to provide intelligent, context-aware responses by AI agents in the new chat session.