your123 - stock.adobe.com
Microsoft intros Azure AI Foundry for building AI apps
The new rebranded platform lets enterprises build generative applications and agents.
While Microsoft's focus at its Ignite developer conference on Tuesday was on Copilot and AI agents, the cloud provider also devoted some efforts to provide an environment for developers to create AI applications.
The tech giant on the first day of the conference introduced the Azure AI Foundry, a platform for developers to design, customize, and manage AI apps and agents. The foundry was formerly known as Azure AI Studio.
"Every application is an AI application, and every new generation of apps has brought a changing set of needs," said CEO Satya Nadella, during his keynote. "AI is transforming how we design and customize and manage apps today."
Azure AI Foundry
The Azure AI Foundry is set to address that change with various capabilities.
One is the Azure AI Foundry SDK, now available in preview. It provides enterprises with tools for customizing, testing, deploying and managing AI apps and agents. The Foundry SDK provides developers with 25 prebuilt app templates, enabling them to integrate Azure AI into their apps.
The cloud provider also re-introduced the Azure AI Foundry portal, which was part of the former Azure AI Studio. Available in preview, the portal is a visual interface that helps developers discover and evaluate AI models, services and tools. The portal includes a new management center and enables teams to manage and optimize AI apps at scale.
The Foundry also boasts Azure AI Agent Service, coming next month in preview. The service will help professional developers orchestrate, deploy and scale enterprise-ready agents to automate business processes. The service includes features such as bring your own storage and private networking so organizations can protect their data.
"The basis for Foundry is for building all the applications, agents, assistants and intelligent applications that will run on it," said Gartner analyst Jason Wong.
Help for enterprises
Azure AI Foundry comes as some see disillusionment with generative AI.
The initial excitement about generative AI tools is now being replaced by a need to move forward and build things with the relatively new technology.
So, Microsoft and competitors, including AWS and Google, are trying to help enterprises that are unsure how to use their technology.
"Enterprises are out there fumbling around," said Mark Beccue, an analyst with TechTarget’s Enterprise Strategy Group. "They don't know what they need. They tend to want a place where they can do integration."
Azure AI Foundry is a starting point enterprises can use to move forward with generative AI technology.
The Foundry is similar to AWS App Studio, introduced in July and made generally available on Nov. 18.
The AWS platform is a generative AI-powered application developer service users can use to create applications with natural language.
Services like Azure AI Foundry and AWS App Studio are foundational steps that will help enterprises build their generative AI journey, Beccue said.
"What I think you're seeing is this evolution towards guidance," he said. "People need more guidance. It's not just ‘do it yourself.’"
AI agents and concerns
Microsoft also introduced ways for users to build AI agents, autonomous and semi-autonomous assistants that can carry out digital tasks with minimal human intervention.
With Microsoft Agent SDK, developers can build agents using Azure AI and Copilot Studio services.
The agents can be deployed across multiple channels, including Microsoft Teams, Microsoft 365 Copilot, the web and third-party messaging platforms.
Agent SDK is available in preview.
The Azure AI Foundry integrations connect Copilot Studio and Azure AI Foundry, enabling agents built in Microsoft Copilot Studio to access the more than 1,800 AI models in the Azure catalog.
With Agent SDK, Microsoft provides users with a framework to control Copilot applications.
"These things are going to be able to make decisions based on data that's held within the company, so it is going to be autonomous," said Futurum Group analyst Keith Kirkpatrick.
However, that approach might not work for some enterprises, Kirkpatrick said.
"If I'm deciding to set up a bunch of agents to do things, there could be some question about, ‘What data can I access? What controls are going to be in place?’" he added.
Enterprises are still concerned about safety and security, especially as agents veer into performing administrative tasks and transactions, Wong, the Gartner analyst, said.
"People want visibility," he said. "They want the ability to have finer grain control over what AI is able to access."
He added that hallucinations could be even more of a problem with AI agents than with standard AI models, which are prone to hallucinations.
"That's something that Microsoft maybe missed an opportunity to really highlight and elevate in terms of these new features and capabilities," Wong continued.
While Microsoft did introduce new governance features across different products and features, such as Power Automate, there is no central governance hub, he said.
Microsoft also introduced Azure AI Content Understanding, a new service for helping developers understand how to build multimodal AI apps.
The cloud provider plans to introduce new fine-tuning options in preview next month in Azure OpenAI Service to enable developers and data scientists to customize models for their business needs.
Partnerships with AI vendors
Microsoft also revealed that it has expanded its partnership with enterprise AI application vendor C3 AI.
The expanded partnership makes Microsoft the preferred cloud provider for C3 AI offerings and establishes C3 as a preferred AI application software provider on Microsoft Azure.
C3 AI's full suite of enterprise AI application software will now be readily available on the Microsoft Commercial Cloud portal.
"This is an inflection point in the global enterprise AI market in terms of indelibly changing the competitive dynamics of this market," C3 AI CEO Tom Siebel said in an interview with TechTarget. He added that joint customers, such as energy giant Shell, want to have providers working together.
"What they want to see is their strategic providers cooperating to make sure their products work together and meet their needs," Siebel continued. "That's what we're committed to doing."
Microsoft is also expanding its partnership with ServiceNow, the self-service and AI platform provider.
The extended partnership includes a collaboration between Microsoft Copilot and ServiceNow AI agents.
The ServiceNow agents aim to address customer problems in Copilot to help with back-end workflows.
Microsoft also previewed its Azure ND GB200 V6 VM, a new AI-optimized virtual machine powered by the Nvidia Blackwell GB200 AI chip.
Esther Ajao is a TechTarget Editorial news writer and podcast host covering artificial intelligence software and systems.
