OpenAI Frontier AI agent platform targets enterprises

AI's 'capability overhang'

Now, the company is moving forward with a combined product and services effort that's precisely as Altman described. The OpenAI Frontier platform will bundle in existing OpenAI tools, such as ChatGPT Enterprise and Business, along with agent-building tools such as the Agents SDK and AgentKit. Frontier projects will also involve forward-deployed engineers – engineers employed by a vendor but embedded within a customer's organization -- who can help integrate these components with existing corporate systems of record and applications, according to an OpenAI spokesperson in an interview with Informa TechTarget Feb. 5.

"The forward-deployed engineers will probably have to work on a more bespoke level to use the existing endpoints that OpenAI has deployed in service of broader goals, like stitching the intelligence layer of Frontier across all their existing infrastructure," the spokesperson said.

As such, OpenAI Frontier isn't a generally available packaged product so far, according to the spokesperson. The company's goal is to work with its six initial customers to match outcomes described in a company blog post Feb. 5, such as a large energy producer that gained $1 billion in additional revenue thanks to an AI agent deployment. That company was not named.

The blog post also pledged enterprise security and governance safeguards, along with the use of open standards to integrate OpenAI Frontier with existing systems.  Specific product details on those components, other than support for the Model Context Protocol project to link AI agents with tools and data, were not yet available.

Instead, what OpenAI brings to the table is its cachet as a frontier model research lab that pioneered generative AI with ChatGPT's seminal launch in 2022, according to the spokesperson.

"We're a research lab that can bring that together with this frontier platform to close what we call the capability overhang," the spokesperson said. "AI models can now be used in a way that applies all the maximum benefits to agents and to whatever apps people want to build on top of their existing architecture."

Another claim staked in the AI gold rush

Meanwhile, OpenAI joins virtually every enterprise IT vendor and cloud hypervisor that has spent the past year making similar claims about their own AI agent platforms, according to Mark Beccue, an analyst at Omdia, a division of Informa TechTarget. Most of these competitors are already better known to enterprise IT buyers than OpenAI, he said.

"All of these other players have large teams of account managers, have established relationships, contracts, NDAs and trust," Beccue said. "It takes years."

OpenAI has been expanding its enterprise sales force, marketing teams and forward-deployed engineers "for quite some time," according to the company spokesperson. That sales force has sold more than 6 million ChatGPT Enterprise seats, among other enterprise products.

However, OpenAI and frontier model developer Anthropic both face pressure to recoup the capital investments they've garnered based on high market valuations, according to Arun Chandrasekaran , an analyst at Gartner. OpenAI is valued at $500 billion and reportedly seeking more financing at a valuation of $800 billion. Anthropic is reportedly discussing a valuation of $350 billion as it seeks another $10 billion in funding. OpenAI last year agreed to the largest private tech funding round on record of $40 billion from SoftBank, while Anthropic's total funding to date is estimated at $37.3 billion.

It's no coincidence that both companies have begun moving beyond their initial focus on frontier AI model development to applications that target businesses, Chandrasekaran  said.

"They have to live up to those valuations, so they have to find new revenue streams," he said. "The truth today is [they] cannot stand alone as model companies. They have to be platform companies, and also application companies."

The argument that OpenAI will make … is that 'If you work with us, you get access to our latest and greatest features faster.'

Facing off against industry heavyweights

This pits OpenAI Frontier against many large competitors, such as Google Cloud, Oracle, Microsoft, AWS and IBM, but there is potential appeal in the company's notoriety in AI model development and the prospect of working closely with expert AI engineers, Chandrasekaran said.

"The argument that OpenAI will make, in my opinion, is that 'If you work with us, you get access to our latest and greatest features faster. … We preview those capabilities that you won't get from the others,'" he said. "The second argument … is you're working with a startup, and more importantly, you're working with a company that's actually building [AI models], and customers will get some very unique access to the CTO or engineering team, and learn from them."

Beccue was skeptical of how successful that pitch will be, especially as big competitors can offer more vertically integrated services, such as native data management tools and database storage, along with AI compute resources.

"The challenge for OpenAI is convincing the market that it can deliver enterprise-grade AI products," Beccue said. "They have largely stuck to a consumer-facing business approach. ... Enterprises tend to worry about things like cost, security and reliability. They have to have a very high level of trust in vendors they use."

Another analyst said he's intrigued by the potential for OpenAI to further refine what models and agents can do with Frontier.

"[What's] interesting is the ambition to let Frontier improve models and agent workflows for consistent operations," said Torsten Volk, also an analyst at Omdia.  "This could include vertical tools and universal agents for troubleshooting, context engineering, and cost control."

Volk said there's great potential appeal to the prospect that OpenAI Frontier can "catch and tame wild agents" created with tools like OpenClaw and Claude Cowork. Both tools were at the center of high-profile security vulnerability reports over the past three weeks.

"These excellent desktop productivity agent frameworks can wreak a ton of havoc at the enterprise level," Volk said. "That's another interesting aspect of the Frontier story."

Beth Pariseau, a senior news writer for Informa TechTarget, is an award-winning veteran of IT journalism. Have a tip? Email her.