Introduction
Okta provides cloud software that manages who — and, increasingly, what — is allowed to log into an organization’s applications, treating identity rather than the network perimeter as the primary point of security enforcement. Todd McKinnon and Frederic Kerrest, who had previously worked together at Salesforce, founded the company in January 2009, initially incorporating it as SaaSure Inc. before renaming it Okta in 2010.
Okta has since become one of the largest independent identity vendors, competing against identity products built into larger platforms such as Microsoft Entra ID. For readers of Brel’s AI in cybersecurity coverage, Okta’s 2025-2026 pivot toward AI agent identity is notable because CEO Todd McKinnon has been unusually direct in public interviews about both the competitive threat AI poses to Okta’s existing business and the scale of the opportunity he believes agent identity represents.
What the company does
Okta’s Workforce Identity Cloud manages employee authentication and access across an organization’s SaaS applications, while its Customer Identity Cloud, built on the Auth0 platform Okta acquired in 2021, gives developers identity and login infrastructure for their own applications. Okta for AI Agents extends this same identity model to a new category of non-human actor: it registers each AI agent as a distinct identity in Okta’s Universal Directory, assigns it a human owner, and enforces least-privilege access using short-lived, scoped tokens rather than hard-coded credentials.
Who it serves
Okta’s customers span large enterprises managing workforce access to hundreds of SaaS applications and developers building customer-facing login flows through Auth0. Okta for AI Agents specifically targets security and platform teams that are deploying autonomous agents across multiple agent-building platforms — Amazon Bedrock, Salesforce Agentforce, ServiceNow, and Anthropic’s Claude among them — and need a single, vendor-neutral place to see and govern all of those agents’ access, regardless of which identity provider a customer already uses for its human workforce.
Company background
Todd McKinnon, a former head of engineering at Salesforce, and Frederic Kerrest, also a Salesforce veteran, founded Okta in San Francisco in January 2009, initially under the name SaaSure Inc. before reincorporating as Okta, Inc. in Delaware in 2010. The company completed its initial public offering on Nasdaq under ticker OKTA on April 7, 2017. In 2021, Okta acquired Auth0, a developer-focused identity platform, for approximately $6.5 billion, substantially broadening its customer base and cap table. McKinnon continues to serve as chairman and CEO, and Kerrest as executive vice chairman, giving Okta continuous founder leadership since its earliest days. The company is headquartered in San Francisco.
Product and AI capabilities
Okta for AI Agents, which reached general availability in April 2026, is built around three capabilities the company describes as a blueprint for agent governance: identity registration, which treats each AI agent as a first-class, non-human identity with a clear human owner rather than an anonymous service account; connection governance, which enforces least-privilege access through short-lived, scoped tokens and monitors agent connections to SaaS applications, APIs, and Model Context Protocol (MCP) servers; and a kill switch, which lets an organization instantly revoke an agent’s access tokens across the enterprise if it exhibits rogue behavior. In May 2026, Okta extended this framework to cover agents built on Amazon Bedrock and, notably, opened the platform to organizations that use a different identity provider — such as Microsoft Entra ID or Ping Identity — for their human workforce, positioning agent identity governance as a distinct, vendor-neutral layer rather than an extension available only to existing Okta customers.
Key developments
Okta went public in 2017 and made its largest acquisition, Auth0, in 2021 to add developer-facing customer identity to its workforce-identity core. Beginning in 2025, CEO Todd McKinnon started speaking publicly and repeatedly about being “paranoid” over what he calls the industry’s “SaaSpocalypse” — the risk that AI coding tools let enterprises simply build their own identity tooling rather than paying for Okta’s. His stated response has been to pivot the company’s R&D toward AI agent identity, culminating in the April 2026 general availability of Okta for AI Agents and its May 2026 expansion to Amazon Bedrock and non-Okta identity providers. In June 2026, Okta said it had become the first independent, neutral identity platform to bring AI agent governance to highly regulated environments.
Why it matters
Okta is a useful case study in how an established SaaS vendor responds in public to an AI-driven threat to its own core business model, rather than only to opportunities AI creates elsewhere. McKinnon’s own framing — that agent identity management could become “the biggest category in cyber by far,” larger than the roughly $28 billion identity market Okta already competes in — gives readers a specific, falsifiable executive prediction to track as enterprises scale up agentic AI deployments over the next several years.
Sector context
Within Brel’s AI in cybersecurity coverage, Okta’s AI agent identity push overlaps with data-security vendors such as Varonis, which governs what data an identity can reach, and with network-security vendors such as Zscaler, which is separately extending zero trust principles to AI agents; each vendor is converging on agent governance from a different layer of the stack — identity for Okta, data access for Varonis, and network traffic for Zscaler. Okta also competes for workforce-identity budgets against platform-embedded identity offerings from much larger vendors, similar to how CrowdStrike competes against Microsoft Defender’s bundled endpoint protection.
Sources and references
This profile draws on Okta’s official newsroom and CEO commentary, alongside independent press coverage.
- SiliconANGLE — “Okta extends AI agent security to Amazon Bedrock…” (2026)
- Okta Newsroom — “Todd’s Take: Securing identity in the age of AI” (2026)
- The Meridiem — “Okta Pivots to AI Agent Identity as ‘Paranoid’ CEO Bets SaaSpocalypse” (2026)
- Wikipedia — “Okta, Inc.”