Microsoft has launched Azure AI Studio at its annual conference, offering businesses the ability to build their own AI “co-pilots” using tools in Azure and machine learning models from their close partner, OpenAI.
Azure AI Studio allows customers to combine a model like OpenAI’s ChatGPT or GPT-4 with their own data, whether it’s text or images, to build a chat assistant or other application that can “reason over” private data.
This new functionality adds to Azure OpenAI Service, Microsoft’s managed product designed to provide businesses with access to OpenAI’s AI lab technologies with additional governance features.
A “co-pilot” is a chatbot application that uses AI, primarily generating text or images, to assist with tasks such as drafting a sales presentation or generating images.
Microsoft has already created several applications of this kind, such as Bing Chat. However, Microsoft’s AI-driven co-pilots cannot leverage a company’s proprietary data to perform tasks, unlike co-pilots created through Azure AI Studio.
The process of building a co-pilot in Azure AI Studio begins by selecting a generative AI model like GPT-4 and assigning it a “meta-prompt” that describes its function and how it should operate.
In addition, AI co-pilots created with Azure AI Studio can incorporate cloud-based storage to keep track of conversations with users and respond with the appropriate context and awareness.
Add-ons extend the capabilities of co-pilots, giving them access to third-party data and other services.
The value proposition of Azure AI Studio lies in allowing customers to leverage OpenAI models on their own data, while complying with their organization’s access policies and rights without compromising security, data policies, or document classification.
Customers can integrate internal or external data they own or have access to, including structured, unstructured, or semi-structured data.
Source: Microsoft
