As AI brokers unfold throughout the enterprise, executives like Todd McKinnon are rethinking how okta agent id can safe each folks and software program in a fast-changing panorama.
Okta, AI disruption, and the SaaSpocalypse concern
Todd McKinnon, co-founder and CEO of Okta, runs a cloud id and safety platform that helps massive firms handle entry throughout apps and companies. With a $14 billion market cap, the agency is a significant software-as-a-service participant. Nonetheless, the trade faces rising stress as AI lets prospects construct their very own instruments as an alternative of paying recurring subscription charges.
This broad concern, generally labeled a “SaaSpocalypse,” has made McKinnon, in his phrases, “paranoid.” On Okta’s most up-to-date earnings name in 2026, he warned that giant language fashions and autonomous brokers might reshape how safety and id merchandise are constructed. That stated, he additionally sees this disruption as an enormous alternative if Okta can transfer rapidly sufficient.
McKinnon says he’s challenge-driven and views AI brokers as a shift probably greater than cloud computing. Furthermore, he believes Okta has already succeeded within the first wave of id companies however should now adapt to seize new markets rising round agent-based architectures.
The increasing marketplace for agent id
Requested why he’s frightened when the full addressable software program market retains rising, McKinnon splits his considerations into two buckets: market shifts and execution. In the marketplace aspect, he argues that AI brokers should log into methods, assume roles, and function with outlined permissions. That creates a wholly new id class alongside conventional human customers.
On the execution aspect, organizations want to soak up extra change throughout expertise stacks, crew buildings, and processes. In his view, firms should elevate their “change quotient” to maintain tempo with agentic methods. Okta due to this fact must orient not simply round workforce id however round changing into an agent id administration platform that permits secure experimentation and adoption of latest instruments.
McKinnon believes the largest alternative is changing into the id layer for AI brokers somewhat than merely defending towards cheaper, vibe-coded rivals. If Okta wins this layer, he says it might develop into the most important class in cybersecurity. Furthermore, with cyber spending already within the a whole lot of billions of {dollars} and id safety a significant slice, the long-term prize is substantial.
Balancing core id with the brand new agent layer
Interrogated about trade-offs, McKinnon rejects a zero-sum framing between conventional id and brokers. He argues that reliability, integration depth, and belief matter in each domains. Even when somebody can recreate particular person options, replicating 1000’s of sturdy integrations and conserving them dependable is extraordinarily troublesome.
He notes that safety and infrastructure software program have traditionally been extra insulated from commoditization, partly as a result of failure has extreme penalties. That stated, he stays vigilant about each aggressive distributors and in-house engineering groups that may attempt to substitute Okta with AI-built id companies.
For McKinnon, the aggressive frontier is evident: enterprises will evolve into “agentic” organizations that rely closely on digital employees. The requirement then turns into centralized management for these brokers, together with a list of the place they run, what methods they connect with, and which permissions they maintain.
OpenClaw, agent dangers, and the necessity for management
The rise of instruments like OpenClaw crystallized these points for Okta. McKinnon calls OpenClaw a watershed second that confirmed the artwork of the attainable for brokers. Nonetheless, it additionally uncovered how troublesome it’s for enterprises to attach the precise knowledge whereas sustaining sturdy safety and governance.
Okta’s response is to construct the rails for secure adoption. These rails embrace an enterprise agent stock system, a system of document for each agent working throughout distributors and platforms. Furthermore, Okta desires to present firms exact controls over which knowledge and functions every agent can entry, plus the flexibility to drag that entry instantly when one thing goes flawed.
When requested if a “kill change” on the agent stage is sufficient, McKinnon emphasizes that detection and response fluctuate by use case. There is no such thing as a magic algorithm that spots all unhealthy habits. As an alternative, Okta is engaged on requirements and indicators that may alert safety groups and revoke entry throughout methods when threshold circumstances are met.
Inside Okta’s construction for the agentic future
Internally, McKinnon says Okta is structured round a people-centric philosophy. Succesful leaders get clear areas of duty, associated features are clustered to cut back communication overhead, and robust administration underpins execution. Furthermore, analysis and improvement is organized by platform to maintain groups targeted on core capabilities.
His management fashion has advanced as the corporate scaled. Early in Okta’s life, decision-making slowed when tasks elevated. Over time, he realized to belief his instincts extra whereas being deliberate about which strategic selections require direct CEO involvement. That stated, he nonetheless dives deeply into matters like agent architectures the place long-term path is at stake.
The strategic choice to pursue okta agent id as a core pillar got here immediately from prospects. Throughout a collection of conferences with lots of Okta’s largest accounts and on the firm’s convention, McKinnon initially pitched a broad unified id platform. As he launched agent ideas, buyer curiosity spiked, and discussions more and more centered on learn how to handle digital employees.
Can right now’s LLMs help the agentic enterprise?
There may be ongoing debate over whether or not present massive language fashions can really help production-scale agent methods. Brokers could be brittle; after they hit boundaries, people should intervene, slowing adoption. McKinnon acknowledges these considerations however insists the expertise will enhance quickly sufficient to justify constructing round it now.
Even with out excessive extrapolations, he believes the market created by agentic workflows will probably be large. Furthermore, he argues that removed from eliminating builders, this future would require extra software program engineering. Groups might want to design architectures, keep methods at scale, and perceive how code generated by brokers behaves over time.
Taking a look at schooling and expertise pipelines, McKinnon expects laptop science fundamentals to stay very important. Nonetheless, curricula will evolve to emphasise coordinating brokers, designing sturdy workflows, and architecting advanced methods. Entry-level engineers, he says, will probably be essential exactly as a result of they’re open to new methods of working.
Information, intelligence, and the place worth accrues
Some analysts declare the database will seize most AI worth as a result of brokers want entry to knowledge. McKinnon takes a extra nuanced stance, distinguishing between uncooked knowledge and intelligence. Clients, he says, need evaluation and insights somewhat than simply storage.
He factors to knowledge platforms like Snowflake, Databricks, and Palantir as examples of methods that already bundle intelligence with knowledge. The open query is who supplies the first intelligence layer in an agentic world: incumbent utility distributors, new specialist gamers, or some mixture of each. Furthermore, he notes that apps have gotten extra related as brokers span a number of silos, which raises demand for requirements governing how brokers entry exterior methods.
This pattern feeds into Okta’s broader imaginative and prescient of an agentic enterprise blueprint. In that mannequin, knowledge, functions, and brokers interoperate by means of standardized id and entry frameworks somewhat than bespoke integrations which can be troublesome to safe.
The blueprint for agent id and management
Okta has outlined a blueprint for the safe agentic enterprise constructed on three pillars: onboarding brokers as a definite id kind, standardizing connection factors, and offering a strong kill change. The primary pillar treats brokers as hybrid identities with attributes of each people and methods.
McKinnon explains that enterprises first want a centralized stock of brokers spanning all distributors and platforms. Some brokers are tightly mapped to particular person workers, others are headless, and a few kind elements of bigger multi-agent workflows. With centralized visibility, organizations can govern which methods every agent can entry and may revoke these hyperlinks when danger exceeds acceptable thresholds.
Designing agent permission and authorization fashions is hard as a result of agent habits is non-deterministic. If firms demand excellent security and deterministic outcomes, they need to tightly prohibit entry. Nonetheless, if they need brokers to be highly effective and autonomous, they need to grant broader permissions and settle for extra danger. Okta’s structure goals to let prospects fine-tune this stability.
Onboarding and modeling brokers
From an architectural perspective, onboarding brokers as id means representing them in id shops with structured attributes, lifecycles, and entitlements. Some patterns require passing a human’s id by means of an agent so downstream methods see actions in a user-centric context. Others require the agent itself to carry its personal id for impartial authorization choices.
This in-between state is what McKinnon means when he says agent id sits between an individual and a system. Furthermore, by modeling brokers explicitly, safety groups can apply insurance policies, audit exercise, and interface with current governance instruments somewhat than treating brokers as invisible background processes.
Detection, indicators, and kill switches
Detecting when an agent does one thing surprising is determined by its meant function and technical implementation. There is no such thing as a common sample that works in each case. As an alternative, Okta is concentrated on defining requirements and sign varieties that functions and safety platforms can use to boost alerts.
The kill change idea is simple: revoke an agent’s entry to methods and knowledge rapidly and constantly. Nonetheless, coordinating that revocation throughout many distributors and clouds requires widespread protocols and integration work, an space the place Okta sees a pure extension of its current strengths.
Platform dynamics, interoperability, and regulation
Requested how main distributors like Salesforce and Microsoft would possibly react if prospects demand cross-silo brokers, McKinnon expects pressure between openness and lock-in. Distributors would possibly attempt to restrict entry or introduce new pricing fashions and charges for cross-platform automation.
Nonetheless, he believes prospects will in the end have leverage, particularly massive enterprises that depend on multi-vendor environments. If lock-in harms prospects, regulators might intervene, simply as they’ve in prior phases of the software program trade’s evolution. Furthermore, as multi-silo agent utilization grows, trade norms and regulatory steering will seemingly adapt to help interoperability.
On the similar time, brokers make it simpler to take away conventional intermediaries in workflows, which might enhance effectivity but in addition create new dependencies on platform insurance policies. This dynamic reinforces the significance of impartial id and entry layers.
Fraud, digital IDs, and AI-powered threats
McKinnon additionally addresses the darker aspect of AI and brokers: scams and fraud. Okta already performs a significant function in authenticating prospects, and AI-powered fraud ways are evolving rapidly. Offline credentials like driver’s licenses and passports are shifting into digital types, together with cellular driver’s licenses and biometric methods.
In his view, these digitized credentials can assist organizations distinguish between human customers, bots, and brokers. Nonetheless, firms additionally must stability fraud prevention with privateness protections and regional laws. Furthermore, combining digital IDs with sturdy id platforms might give enterprises new instruments towards AI-driven assaults.
On age verification and youth security, McKinnon says governments are actively debating digital ID requirements and necessities. Okta’s function is to make sure prospects can settle for and implement no matter regulatory frameworks emerge whereas sustaining sturdy safety and suppleness.
What comes subsequent for Okta and agent id
Wanting forward, McKinnon frames Okta’s mission as serving to enterprises construct safe agentic environments. The corporate is advancing its blueprint and transport instruments that permit prospects onboard brokers safely, standardize connections, and train controls similar to kill switches when danger surfaces.
Furthermore, he argues that as brokers turn out to be pervasive, id platforms should prolong seamlessly from people to software program. In that imaginative and prescient, the identical core rules that secured workforce logins will govern digital employees, unleashing productiveness whereas preserving safety, compliance, and organizational belief.
In abstract, Okta is betting that the way forward for safety will hinge on rigorous administration of agent identities, giving enterprises a unified solution to see, management, and audit each human and non-human actor throughout their methods.