OpenAI Exec Is Still ‘Bullish’ About the Future of Software

OpenAI has a pitch for software companies freaking out about AI and this week’s massive sell-off: Work with us, not against us.

On Thursday, OpenAI announced “Frontier,” its new enterprise-focused platform that is set to build on the startup’s existing ChatGPT enterprise offerings and its growing team of Forward Deployed Engineers. The product is designed to give agents greater autonomy across multiple systems rather than performing isolated tasks.

“Are software companies going to need to adapt to AI in general? Yes, of course,” Fidji Simo, OpenAI’s CEO of Applications, told reporters during a call ahead of Thursday’s announcement. “But I think the ones that do adapt to AI will actually find Frontier to be a massive opportunity for them.”

It’s been a rough week for software

Wall Street swung against software companies earlier this week after Anthropic released a tool that can do clerical work for legal professionals. In tech, there is no shortage of voices pronouncing the end of SaaS.

Simo said OpenAI’s new offering, which already counts Uber, State Farm, Thermo Fisher, and Intuit as early adopters, isn’t another shot at existing software companies.

“We’re not going to build every single AI agent that companies need — absolutely not,” Simo said, “And that’s why we have built a platform in a way where the third-party software companies can deploy their agents on top of us.”

Frontier’s design, Simo said, was informed by her conversations with software companies before she formally joined OpenAI’s leadership team in May. She said software companies asked whether OpenAI could offer a platform that simplifies integration, making it easier for them to deploy agents through OpenAI’s enterprise relationships.

Barret Zoph, OpenAI’s GM of B2B, said the goal of Frontier is to help companies get the full value out of current models, which is difficult to do with one-size-fits-all offerings.

“What we’re fundamentally doing is transitioning agents into true AI coworkers,” he said.

OpenAI is one of many companies racing to roll out agent-based systems. Last month, Anthropic released an agent-based tool that allows AI to take on multi-step tasks that could automate some white-collar work. Meanwhile, a former xAI engineer said in a recent podcast that Elon Musk’s company was developing and testing human emulators to automate white-collar work.

OpenAI declined to share how much revenue it could generate with the new platform. The company has a major need for additional revenue streams, with roughly $1.4 trillion in data center projects committed over the next year. On top of that, OpenAI recently announced plans to bring ads to lower tiers of ChatGPT.

OpenAI said Frontier is available to “a limited set of customers” and will have more availability coming soon. The company declined to discuss the cost of the service.

The company said Frontier is built on open standards and is designed to work with agents developed in-house or by third-party vendors, including competitors.