Asserting the general public preview of Deep Analysis in Azure AI Foundry—an API and SDK-based providing of OpenAI’s superior agentic analysis functionality.
Unlock enterprise-scale net analysis automation
Immediately we’re excited to announce the general public preview of Deep Analysis in Azure AI Foundry—an API and software program improvement equipment (SDK)-based providing of OpenAI’s superior agentic analysis functionality, totally built-in with Azure’s enterprise-grade agentic platform.
With Deep Analysis, builders can construct brokers that deeply plan, analyze, and synthesize info from throughout the online—automate complicated analysis duties, generate clear, auditable outputs, and seamlessly compose multi-step workflows with different instruments and brokers in Azure AI Foundry.
AI brokers and data work: Assembly the following frontier of analysis automation
Generative AI and huge language fashions have made analysis and evaluation quicker than ever, powering options like ChatGPT Deep Analysis and Researcher in Microsoft 365 Copilot for people and groups. These instruments are remodeling on a regular basis productiveness and doc workflows for hundreds of thousands of customers.
As organizations look to take the following step—integrating deep analysis immediately into their enterprise apps, automating multi-step processes, and governing data at enterprise scale—the necessity for programmable, composable, and auditable analysis automation turns into clear.
That is the place Azure AI Foundry and Deep Analysis are available: providing the flexibleness to embed, prolong, and orchestrate world-class analysis as a service throughout your complete enterprise ecosystem—and join it together with your information and your programs.
Deep Analysis capabilities in Azure AI Foundry Agent Service
Deep Analysis in Foundry Agent Service is constructed for builders who wish to transfer past the chat window. By providing Deep Analysis as a composable agent software by way of API and SDK, Azure AI Foundry allows clients to:
- Automate web-scale analysis utilizing a best-in-class analysis mannequin grounded with Bing Search, with each perception traceable and source-backed.
- Programmatically construct brokers that may be invoked by apps, workflows, or different brokers—turning deep analysis right into a reusable, production-ready service.
- Orchestrate complicated workflows: Compose Deep Analysis brokers with Logic Apps, Azure Capabilities, and different Foundry Agent Service connectors to automate reporting, notifications, and extra.
- Guarantee enterprise governance: With Azure AI Foundry’s safety, compliance, and observability, clients get full management and transparency over how analysis is run and used.
In contrast to packaged chat assistants, Deep Analysis in Foundry Agent Service can evolve together with your wants—prepared for automation, extensibility, and integration with future inner information sources as we broaden help.
The way it works: Structure and agent circulate
Deep Analysis in Foundry Agent Service is architected for flexibility, transparency, and composability—so you’ll be able to automate analysis that’s as strong as your small business calls for.
At its core, the Deep Analysis mannequin, o3-deep-research, orchestrates a multi-step analysis pipeline that’s tightly built-in with Grounding with Bing Search and leverages the newest OpenAI fashions:
- Clarifying intent and scoping the duty:
When a consumer or downstream app submits a analysis question, the agent makes use of GPT-series fashions together with GPT-4o and GPT-4.1 to make clear the query, collect further context if wanted, and exactly scope the analysis process. This ensures the agent’s output is each related and actionable, and that each search is optimized for your small business situation. - Net grounding with Bing Search:
As soon as the duty is scoped, the agent securely invokes the Grounding with Bing Search software to collect a curated set of high-quality, latest net information. This ensures the analysis mannequin is working from a basis of authoritative, up-to-date sources—no hallucinations from stale or irrelevant content material. - Deep Analysis process execution:
The o3-deep-research mannequin begins the analysis process execution. This entails pondering, analyzing, and synthesizing info throughout all found sources. In contrast to easy summarization, it causes step-by-step, pivots because it encounters new insights, and composes a complete reply that’s delicate to nuance, ambiguity, and rising patterns within the information. - Transparency, security, and compliance:
The ultimate output is a structured report that paperwork not solely the reply, but in addition the mannequin’s reasoning path, supply citations, and any clarifications requested through the session. This makes each reply totally auditable—a must have for regulated industries and high-stakes use circumstances. - Programmatic integration and composition:
By exposing Deep Analysis as an API, Azure AI Foundry empowers you to invoke analysis from anyplace—customized enterprise apps, inner portals, workflow automation instruments, or as half of a bigger agent ecosystem. For instance, you’ll be able to set off a analysis agent as a part of a multi-agent chain: one agent performs deep net evaluation, one other generates a slide deck with Azure Capabilities, whereas a 3rd emails the outcome to choice makers with Azure Logic Apps. This composability is the actual game-changer: analysis is now not a handbook, one-off process, however a constructing block for digital transformation and steady intelligence.
This versatile structure means Deep Analysis might be seamlessly embedded into a variety of enterprise workflows and purposes. Already, organizations throughout industries are evaluating how these programmable analysis brokers can streamline high-value eventualities—from market evaluation and aggressive intelligence, to large-scale analytics and regulatory reporting.
Pricing for Deep Analysis (mannequin: o3-deep-research) is as follows:
- Enter: $10.00 per 1M tokens.
- Cached Enter: $2.50 per 1M tokens.
- Output: $40.00 per 1M tokens.
Search context tokens are charged enter token costs for the mannequin getting used. You’ll individually incur prices for Grounding with Bing Search and the bottom GPT mannequin getting used for clarifying questions.
Get began with Deep Analysis
Deep Analysis is out there now in restricted public preview for Azure AI Foundry Agent Service clients. To get began:
We will’t wait to see the modern options you’ll construct. Keep tuned for buyer tales, new options, and future enhancements that may proceed to unlock the following era of enterprise AI brokers.
