San Francisco, February 5, 2026, 07:26 PST
- OpenAI rolls out Frontier to help firms build and manage “AI agents” that can carry out tasks
- Company is pitching Frontier as an enterprise layer that can also work with third-party agents
- Pricing was not disclosed; rollout starts with a limited set of customers
OpenAI on Thursday launched Frontier, a service for companies to build and manage artificial-intelligence agents — software tools that can carry out tasks such as fixing a software bug — as it steps up its push to sell to businesses. Fidji Simo, OpenAI’s CEO of applications, said the platform is meant to be an “intelligence layer” to help enterprises deploy agents more easily. (Reuters)
The timing reflects a shift inside big companies: pilots and demos are giving way to systems that touch real data and real workflows. That brings new questions fast — who the software is allowed to act for, what it can access, and how managers track mistakes.
Agents have become the next battleground because they promise automation, not just answers. If a tool can read a request, plan steps, and take an action inside a company system, it can cut time — or create a mess at high speed.
OpenAI has been describing Frontier as a way to make agents usable across the messy reality of corporate tech stacks, where data sits in different clouds and apps. The company’s bet is that a single management layer can make adoption less painful, and lock in longer-term enterprise spending.
In a company blog post, OpenAI said Frontier can connect internal systems such as data warehouses, customer-relationship management tools and ticketing platforms to give agents shared business context, while adding identity, permissions and guardrails for sensitive work. OpenAI also listed early adopters including HP, Intuit, Oracle, State Farm, Thermo Fisher and Uber, and said pilots have included BBVA, Cisco and T-Mobile. State Farm executive vice president Joe Park said: “Partnering with OpenAI helps us give thousands of State Farm agents and employees better tools to serve our customers.” (OpenAI)
Frontier is available on Thursday to a limited set of customers, with broader availability expected over the next few months, OpenAI said. Chief revenue officer Denise Dresser declined to disclose pricing in a press briefing, and OpenAI business-to-business chief Barret Zoph called Frontier an “agent interface” meant to sit above fragmented tools and siloed data. The push lands as OpenAI faces enterprise competition from Microsoft’s Agent 365 and from Anthropic, which has been expanding its own agent offerings. (The Verge)
OpenAI is also leaning on flexibility as a selling point, saying Frontier is built to work with a company’s existing setup rather than forcing a replatform. That matters to big firms that already have in-house models, vendor contracts, and compliance policies they do not want to rebuild from scratch.
For buyers, the appeal is control: one place to set what an agent can see, what it can do, and where it is allowed to run. For OpenAI, the appeal is simpler — enterprise budgets tend to be sticky once a platform becomes infrastructure.
But the downside risk is baked into the pitch. Companies still have to decide how much autonomy to give systems that can act on their own, and an agent that writes code, moves money, or updates customer records can cause damage when it fails, even if the failure is rare. The lack of disclosed pricing adds another uncertainty for procurement teams.
Frontier’s rollout over the next few months will show whether businesses want a neutral control plane for fleets of agents — or whether they keep buying narrow tools tied to specific vendors and tasks.
