When OpenAI launched Frontier in February, the announcement was described as a platform for enterprise AI brokers. What it really signalled was a problem to the income structure underpinning the software program business.
Frontier is designed to behave as a semantic layer in an organisation’s current techniques, connecting information warehouses, CRM platforms, ticketing instruments, and inner functions so AI brokers can function with the identical enterprise context a human worker would have. OpenAI describes these brokers as “AI coworkers” that may be on-boarded, assigned identities, granted permissions, and reviewed for efficiency.
Early prospects embrace Uber, State Farm, Intuit, and Thermo Fisher Scientific. OpenAI CFO Sarah Friar has said that enterprise prospects presently account for roughly 40% of the corporate’s income, and she or he goals to extend this determine to nearer to 50% by year-end, and claims Frontier is the automobile.
Frontier in enterprise workflows
The case for Frontier is that brokers deployed in isolation add complexity not take away it. Every new agent is some extent of integration, requiring its personal information connections and governance controls, and the result’s fragmentation. OpenAI’s reply is a shared enterprise context. Fairly than every agent constructing its personal understanding of how an organisation works, Frontier supplies a centralised layer that every one brokers can reference.
Fidji Simo, OpenAI’s CEO of Functions, talking on the launch briefing, referred to her time operating Instacart. “We spent months integrating every of those that we chosen. We didn’t even get what we really needed, as a result of every device was good for one use case, however they weren’t built-in or speaking to at least one one other, so we have been simply reinforcing silos on silos.”
The outcomes OpenAI cites from early deployments embrace a world funding agency utilizing Frontier brokers in its gross sales course of releasing up greater than 90% of salesperson time beforehand spent on administrative duties. A know-how buyer reported saving 1,500 hours a month in product growth. At a significant producer, brokers compressed a manufacturing optimisation course of from six weeks to a single day.
Frontier manages brokers constructed by OpenAI, in-house enterprise groups, and people from third-party suppliers. Openness is a design precept and positioning: it makes Frontier tougher to dismiss on the grounds of vendor lock-in and expands the floor space it may well govern.
The seat-licence downside
A deep concern for incumbents is structural. The per-seat licence mannequin that has made SaaS enormously worthwhile assumes that software program use maps to headcount. If an AI agent handles the workflow that beforehand required a human worker logging into Salesforce, the justification for that seat licence weakens. Fortune described worry available in the market of fashions like Frontier making SaaS software program “invisible” and consequently much less precious.
Salesforce’s inventory has declined greater than 27% this yr, which analysts have attributed extra to agentic AI disruption fears than to any weak point in its underlying financials. Income reached $11.2 billion within the quarter, Agentforce’s annual recurring income hit $800 million, and the corporate closed 29,000 Agentforce offers. Inventory fell after steerage that got here in under Wall Avenue’s expectations.
The incumbents will not be standing nonetheless. Salesforce has launched what it calls the Agentic Enterprise License Settlement, a fixed-price, all-you-can-eat mannequin for Agentforce that makes an attempt to make consumption extra predictable for enterprise patrons.
ServiceNow has moved to consumption-based pricing for a few of its AI agent choices, and in January signed a multiyear settlement with OpenAI to embed frontier mannequin talents straight into its platform. Microsoft has launched consumption-based pricing with its per-user mannequin for Copilot Studio.
The pricing pivot alerts that firms perceive the seat-licence mannequin can not survive agentic AI unchanged. The query is whether or not repricing is sufficient or whether or not the structure itself wants to vary.
Two concepts of the place the intelligence layer ought to sit
Ought to AI brokers reside inside techniques of file, or above them? Salesforce and ServiceNow are betting on the embedded mannequin, arguing that brokers are only once they sit closest to the info, and that CIOs will belief governance and compliance controls extra readily from distributors already managing their workflows.
Marc Benioff, CEO of Salesforce, has described Agentforce because the “working system for the agentic enterprise.” ServiceNow positions its AI Management Tower as a centralised governance layer for all brokers, no matter origin.
OpenAI, and to the same diploma, Anthropic with Claude Cowork, is betting on the overlay mannequin. Frontier sits above current techniques, utilizing open requirements to attach them. The pitch is that enterprises shouldn’t need to re-platform to get production-grade brokers operating of their operations.
Each arguments have benefit, and enterprises evaluating these platforms will discover real trade-offs. The embedded method provides tighter information management and quicker time to worth in a recognized ecosystem. The overlay method provides flexibility and avoids the issue of brokers that may solely see one vendor’s information.
What the incumbents have that OpenAI doesn’t is a long time of institutional belief and current contracts. What the AI chief has is the mannequin capacity benefit and an argument that it may well run the intelligence layer throughout the entire enterprise.
Frontier is presently accessible to a restricted set of consumers, with broader availability anticipated over the approaching months. Pricing has not been disclosed, with OpenAI directing organisations to its enterprise gross sales crew.
Many giant enterprises run Salesforce, ServiceNow, and Microsoft infrastructure concurrently. The fast query is whether or not Frontier turns into an orchestration layer that connects techniques, or a platform that displaces them.
OpenAI’s chief income officer, Denise Dresser, mentioned: “What’s actually lacking nonetheless for many firms is only a easy option to free the facility of brokers as teammates that may function contained in the enterprise with out the necessity to rework every little thing beneath.”
Each platform on this area claims to shut the hole. SaaS incumbents have a head begin on belief and information. Whether or not that proves enough is the central query for enterprise software program via to the tip of 2026.
(Picture by Austin Distel)
See additionally: OpenAI’s enterprise push: The hidden story behind AI’s gross sales race
Need to study extra about AI and massive information from business leaders? Take a look at AI & Big Data Expo happening in Amsterdam, California, and London. The great occasion is a part of TechEx and is co-located with different main know-how occasions together with the Cyber Security & Cloud Expo. Click on here for extra data.
AI Information is powered by TechForge Media. Discover different upcoming enterprise know-how occasions and webinars here.
