AI "colleagues" about to start work? OpenAI launches Frontier, targeting enterprise-level agent automation
```
OpenAI announced on Thursday the launch of a brand-new platform designed to help enterprises deploy AI agents more easily. According to media reports, this is part of the company’s broader strategy, aimed at consolidating its leading position in automating high-value work tasks.
This new product, called Frontier, allows enterprises to build and manage AI tools, ensuring every AI agent has appropriate safety guardrails and data access permissions. The platform’s goal is to simplify the deployment process for AI agents and pave the way for broader adoption of this technology among corporate clients.
OpenAI stated that Frontier can be used in conjunction with the AI agent building tools previously released by the company, making it easier for enterprises to integrate various data sources required for AI agents to perform tasks. With this platform, AI agents can handle information from different systems and carry out tasks such as processing files and running code.
Media reports indicate that top AI companies, including OpenAI and Anthropic, are increasingly betting on AI agents. These tools are designed to perform tasks on behalf of individuals with minimal human intervention. This trend has recently caused significant market volatility, as investors worry these tools could impact traditional software vendors. Meanwhile, these AI startups are accelerating their race to win enterprise clients across industries to boost revenue and offset the immense costs of developing more advanced AI software.
During a phone briefing with the media, OpenAI’s head of applied business, Fidji Simo, described these agents as “AI colleagues.” She said these AI agents can not only collaborate with humans but also work alongside AI agents developed by OpenAI competitors such as Anthropic and Microsoft.
“We are systematizing the experience we’ve gained working directly with hundreds of large enterprises—how to build, deploy, and manage AI agents and truly get work done at scale—to make the whole process much simpler.”
“By the end of this year, most digital work in leading enterprises will be directed by humans and carried out by swarms of AI agents. This is already happening in the programming field and will expand to more domains.”
Notably, OpenAI executives said the platform will allow enterprises to access some competitor AI agent software, including Anthropic. This means Frontier could potentially become a one-stop platform for enterprise use of AI agents. In the past, Anthropic has at times restricted competitors’ access to its underlying technology.
OpenAI stated that companies such as Intuit, State Farm, and Thermo Fisher Scientific are already using Frontier, and dozens of OpenAI clients are also testing the product. OpenAI has also partnered with companies like ServiceNow to integrate its AI model directly into the business software company’s AI agents. However, OpenAI has not yet disclosed the pricing for Frontier.
The Goal is Not to Replace Existing Software Tools
OpenAI is making this announcement at a time when software stocks saw significant declines on Tuesday afternoon, with shares of companies from PayPal to Expedia and Intuit dropping more than 10%. Software and data-related stocks lost over $300 billion in market value. The market reaction mainly stemmed from investor worries that AI-driven disruptive technology will reduce demand for traditional software tools.
Earlier this week, OpenAI and Anthropic both released new products, believed to be a major trigger for the recent sell-off in software stocks. Anthropic recently expanded its Claude-powered Cowork assistant to include a range of plugins capable of executing professional business tasks, including plugins for the legal industry.
On the same day as the release of Frontier, Anthropic launched a new AI model, Claude Opus4.6, specifically for conducting financial research. The company said this model can review company data, regulatory filings, and market information to generate detailed financial analysis—tasks that usually take humans several days to complete. Meanwhile, OpenAI also released a new version of its programming tool Codex on Thursday, whose operation is similar to the applications Anthropic is building for Claude.
By contrast, Simo said the launch of Frontier is "excellent news" for the software industry, as the platform is not intended to replace existing software tools. She said Frontier's positioning is to help enterprises distribute and run their own AI agents.
Simo said:
“We cannot build every AI agent that enterprises need. That is exactly why we designed the platform for all software companies to deploy AI agents on it.”
Participants may include OpenAI’s backer Microsoft, as well as companies like Oracle and SAP. These enterprises themselves offer specialized AI agents for automating business processes. OpenAI stated that these companies can use Frontier to promote and deploy their own AI agents and serve as sources of business data to support the operation of customized AI agents.
For OpenAI’s AI agents to function in certain scenarios, they need access to customer data in CRM systems like Salesforce and content from instant messaging apps like Slack, Simo said.
Media reports said the launch of Frontier is also intended to help OpenAI attract more enterprise users as it competes with Anthropic, Google, and other rivals for corporate clients. By making Frontier a "de facto standard" for building and managing AI agents—regardless of whether these agents are developed by OpenAI—the company’s goal is to bring more enterprise clients into its overall AI ecosystem.
Risk Warning and DisclaimerThe market carries risks, and investments should be made with caution. This article does not constitute personal investment advice, nor does it take into account individual users' special investment goals, financial situations, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular circumstances. Investing based on this article is at your own risk. ```