OpenAI released Frontier on Thursday, a new artificial intelligence platform that can help companies build, deploy and supervise AI agents. The AI company said Frontier works with OpenAI's previously released AI agent building tools to make it easier for companies to integrate the data sources needed for agents to perform tasks. According to OpenAI, these agents will be able to process information from a variety of sources and complete tasks such as processing files and running code.
In a conference call with reporters, OpenAI application business CEO Fei Ji. Fidji Simo describes these agents as "AI colleagues" that can collaborate with humans and can be used alongside agents developed by OpenAI competitors such as Anthropic and Microsoft.
"By the end of this year, the majority of digital work in leading enterprises will be directed by humans and performed by legions of agents," Simo said. "This is already true in coding and will be true in many other fields as well."
The company did not disclose how much Frontier charges, and the platform is currently only available to some customers.
The move comes after software stocks ranging from PayPal to Expedia and Intuit plunged more than 10% on Tuesday afternoon, wiping more than $300 billion off the market value of software and data stocks. The stock price plunge was largely driven by concerns that AI-driven disruption will reduce demand for traditional software tools.
Earlier this week, OpenAI and Anthropic both released new products, triggering a plunge in software stocks. Anthropic recently expanded the capabilities of its Claude-powered Cowork assistant, adding new plugins to perform specialized business functions, including one for the legal industry. OpenAI released a new version of its coding tool Codex on Monday that works similarly to the app Anthropic is integrating into Claude.
In contrast, OpenAI's Simo said Frontier's release is "excellent news for the software industry" because it is not designed to replace existing software tools. Frontier will serve as a way for enterprises to deploy their own AI agents, she said.
"We're not going to build every AI agent that a company needs," Simo said. "Because of that, we're building the platform in a way that all these software companies can deploy their agents on top of our platform."
These players may include OpenAI backers Microsoft, Oracle and SAP - all of which offer their own AI agents specifically designed to automate business processes. OpenAI said these companies can use OpenAI's Frontier to get their agents adopted and serve as a source of business data needed to customize the agents to run.
In some cases, for OpenAI's agents to work, they need to get customer data from customer relationship management systems like Salesforce and content from messaging apps like Slack, Simo said.
OpenAI has also struck deals with companies such as ServiceNow to integrate its AI models directly into the commercial software developer's AI agents.
Frontier's first customers include Intuit, State Farm, Thermo Fisher and Uber, the company said. Dozens more OpenAI customers are testing the product.
The release of Frontier is also aimed at helping OpenAI attract more enterprise customers, for which it is competing with Anthropic, Google and other rivals. By establishing Frontier as a standard for building and managing AI agents, whether developed by OpenAI or not, the company aims to bring more enterprise customers into its overall AI ecosystem.
As previously reported, the San Francisco-based AI laboratory is preparing to go public in the fourth quarter of this year.
Parent company News Corp has a content licensing partnership with OpenAI.