OpenAI Frontier and General Bots 🟡 BETA

OpenAI Frontier is the enterprise platform OpenAI launched in February 2026 to build, deploy and govern AI agents across business systems. It occupies the enterprise end of the agent market, which is the same ground General Bots targets — from a very different starting point.

Verified 2026-09-13. Sources: OpenAI’s Frontier launch material (openai.com/business/frontier) and published coverage of the launch. Where a capability is not publicly documented, this page says so rather than guessing. Re-check before quoting.

What OpenAI Frontier is

AspectDetail
What it isAn enterprise platform for building, deploying and managing AI agents across business systems
Launched5 February 2026
Core conceptsBusiness Context (a shared institutional-memory layer over data warehouses, CRM, internal apps and documents), Agent Execution (parallel, multi-step tasks with escalation to humans), Open Platform (manages agents built outside OpenAI too)
GovernanceExplicit permissions, comprehensive auditing, and auditable actions for every agent interaction
Model policyNot locked to OpenAI models; multi-vendor agent management is a deliberate design goal
Management modelTreats agents like employees: onboarding, feedback loops and improvement cycles
AvailabilityEnterprise contracts; not self-serve signup
Launch customersHP, Intuit, Oracle, State Farm, Thermo Fisher and Uber, with BBVA, Cisco and T-Mobile in pilots

Frontier is explicitly not a chat interface (that is ChatGPT), not a low-code builder, and not something you sign up for yourself. It is a management and governance layer.

Where they overlap

  • Agent governance as a product. Both treat permissions, auditing and accountability as core features rather than afterthoughts.
  • Multi-step agent execution. Both run agents across systems rather than answering single questions.
  • Multi-vendor models. Neither insists you use one model provider.
  • Systems of record. Both connect to the CRM, databases and document stores you already run.

Where they differ

AspectOpenAI FrontierGeneral Bots
DeploymentManaged cloud, enterprise contractSelf-hosted on your own infrastructure
LicensingProprietary, enterprise agreementMIT open source
Data locationOpenAI-hostedYour servers; you choose residency
Self-serveNo — enterprise onboardingYes — install and run it yourself
ScopeAgent management layerAgents plus a full productivity suite (mail, calendar, files, tasks, meetings, office apps)
InterfaceAdministration and governance surfacesWeb chat, messaging channels, voice and a desktop suite
ConfigurationAPIs and SDKsBASIC scripting, plus conversational configuration
Cost modelEnterprise contractInfrastructure plus your own LLM spend
MaturityBacked by a major vendor, Fortune 500 deploymentsEarly preview; stable surface is Chat, Explorer and Vibe

When Frontier is the better fit

  • You are a large enterprise that needs a vendor-supported agent management layer with contractual governance.
  • Your agents must be centrally governed across many business units and systems of record.
  • You want to manage agents built on several vendors’ models from one control plane.
  • You have no appetite to run infrastructure yourself.

When General Bots is the better fit

  • You need data sovereignty — the agents and the data they touch must stay on your own infrastructure.
  • You want open source you can read, modify and audit, with no per-user licensing.
  • You want the agent and the daily productivity tools in one deployable system, not an agent layer bolted onto tools hosted elsewhere.
  • You need channels your customers actually use — WhatsApp, Telegram, SMS, email and voice — as first-class surfaces.

An honest note on scale

Frontier launched with Fortune 500 customers and a major vendor behind it. General Bots is an early-preview project maintained by a small team. If you need enterprise support contracts today, Frontier is the safer choice; if you need to own the stack, General Bots is the one that lets you.

Summary

Frontier answers “how does a large enterprise govern many agents across many systems?” General Bots answers “how does an organisation own its assistant and its data end to end?” Both are legitimate questions; they are not the same question.

See Also