The Identity Fabric and the intent-driven approach are the way to propel it.
I came back from Identiverse 2026 more excited about the opportunity in front of us than at any point since we started building aizome.
Not because of the security conversations - although those were substantive and the identity community (across CISOs, CIOs, and IAM teams) holds a wide consensus that the Identity-led approach is the superior way to control AI Agents, but because of what is happening in enterprises right now, in the business units and operations floors and finance departments of every large organization on the planet.
Enterprise AI agents are already revolutionizing how work gets done. And the organizations that figure out how to trust them - fully, confidently, at scale - are going to pull away from the ones that don't in ways that will be very difficult to close.
That is the opportunity I want to talk about. Not the risk. The opportunity.
What's Actually Happening Out There
The numbers from 2026 tell a story that most security conversations understate.
79% of organizations are already using AI agents to some degree, with 88% planning to increase their budgets specifically for agentic capabilities. 66% report measurable productivity improvements, with 62% expecting ROI exceeding 100%. - PWC AI Agent Survey
Knowledge workers using production AI agents recover a median of 6.4 hours per week per seat. Customer service AI agents resolve a contained ticket for $0.46 versus $4.18 human-handled. Code review agents complete a routine PR for $0.72 versus $48 in senior engineer time.
Gartner projects that by the end of 2026, 40% of enterprise applications will include task-specific AI agents - up from less than 5% in 2025. In a best-case scenario, agentic AI could generate nearly 30% of enterprise application software revenue by 2035, surpassing $450 billion.
This is not a future trend. This is happening right now, in production, at scale, in every industry.
Finance agents are automating the month-end close and vendor payment processing. HR agents are managing onboarding workflows and policy queries across thousands of employees simultaneously. Sales agents enrich CRM data, qualify leads, and prepare account briefings in real time. Operations agents monitor supply chains, flagging anomalies and initiating procurement workflows without human intervention. Customer service agents resolve complex issues end-to-end without escalation.
The productivity gains are real. The business value is proven. The ROI is compelling. And the organizations deploying these agents are not slowing down.
The Problem Is Not the Agent. It Is the Trust Gap.
Here is the observation I keep coming back to, and the one I made on stage at the Non-Human Identity & Agentic AI Summit on Monday:
Only 29% of organizations see significant ROI from generative AI, and just 23% from AI agents. The individual wins are real and measurable, but they are not translating to business value at the organizational level.
Why? Because most organizations are deploying AI agents at the edges of their operations, in bounded, low-stakes contexts, with human oversight for every significant action, in environments where the consequence of a mistake is limited. They are not yet deploying agents into the core of their business: the sensitive cross-functional workflows, the financial systems, the HR processes, the mission-critical operations where the real productivity multiplier lives.
And the reason they are not is not that the technology isn't ready. It is because they cannot trust the agents enough to delegate meaningful work to them and be accountable.
Trust, in this context, is not a sentiment. It is an architectural property. You trust an agent when you can see everything it does, understand why it does it, verify that its behavior and intent is consistent with its purpose, and hold someone accountable when it isn't. Trust is what you give a new employee after they have demonstrated that they operate within the boundaries of their role, that their judgment aligns with the organization's values, and that their actions are transparent and auditable.
We hold every employee accountable for their actions. AI agents should be no different. And when they are, when every agent has an identity, a boundary, an owner, and an observable intent, you can delegate to them with the same confidence you would delegate to a trusted member of your team.
That is the unlock. Not restricting AI. Earning the right to trust it.
The ARISE Category: The Identity Fabric Infrastructure That Makes Trust Possible
The analyst community is beginning to name what organizations need to close the trust gap.
SACR recently identified a new emerging category they call ARISE: Agentic Runtime Identity Security Enforcement. Their framing captures the market signal precisely. Recent acquisitions by SailPoint, Cisco, and 1Password all point to the same convergence: identity security is becoming the de-facto control layer for the agentic enterprise.
The ARISE category is built on a foundational insight: you cannot trust what you cannot see, and you cannot govern what you cannot identify. The Identity Fabric infrastructure that makes enterprise AI agents trustworthy operates across four domains:
Discovery - finding every agent operating in your environment, including the ones no one approved and the ones built by employees, solving problems in workflow tools and on their laptops. You cannot delegate meaningful work to an agent population you have not mapped.
Identity - giving every agent a verified identity, a documented owner, and a defined scope. Managing the entire Identity lifecycle across Joiner, Mover, and Leaver. Not because identity alone provides governance, but because accountability requires a clear chain of authority from agent action back to human authorization. Without identity, trust is impossible.
Intent governance - the capability that most organizations are still building toward and that I believe is the most important unlock for enterprise AI - thus intent-driven is the very foundation of the aizome platform. An agent's intent is its reasoning, what it believes it is supposed to be doing, why, and for whom. Steven Rennick, highlighted it nicely in his panel at the NHAI Summit. Governing intent means continuously observing whether an agent's behavior is consistent with the purpose it was given, detecting when it drifts, and responding before the drift produces an outcome nobody authorized.
This is the same standard we hold human employees to. A Finance analyst has a job description. A trusted employee operates within it. When their behavior diverges from what they were hired to do, there is a process for catching it, understanding it, and addressing it. Intent governance for AI agents is the same principle applied at machine speed.
Accountability - the tamper-evident audit trail that proves every agent action back to the human authorization that permitted it. Not just for compliance. For the business confidence that lets you delegate more.
aizome is built to provide all four, geared specifically towards the enterprise AI agents operating in core workforce business processes, where the stakes are highest, and the governance gap is widest.
What the Organizations Winning with AI Have in Common
What separates the organizations seeing real ROI from the majority? They share four characteristics: they tie AI directly to revenue outcomes, they architect platforms that give business teams autonomy while IT retains oversight, they build governance infrastructure concurrent with deployment, and they treat AI agents as trusted collaborators rather than experimental tools.
That last point is the one I keep returning to. Trusted collaborators. Not tools to be monitored from a distance. Not experiments to be run in sandboxes. Collaborators to be given real work, real responsibility, and real accountability.
The enterprises that get there first will have an advantage that is very difficult to replicate. Not a first-mover advantage in the technology, the models and frameworks are accessible to everyone. A governance advantage. The infrastructure, the processes, and the organizational confidence to deploy AI agents into the core of their operations, while their competitors are still running pilots at the edges.
Autonomous agents and agentic AI surged 31.5% year-over-year as a top technology priority in 2026. The productivity argument has matured; enterprises are now demanding that every AI capability connect directly to revenue growth or margin improvement.
The conversation has moved from "can AI agents help?" to "how do we deploy them in ways that transform the business?" The answer to that question is governance infrastructure. Not as a constraint on AI adoption, but as the enabler of it.
The Next 24 Months
I made a prediction on the Identiverse panel that I want to reframe here, because the framing matters. I said that within 24 months, every enterprise will have experienced a material security incident attributable to an AI agent they did not know was running in their environment. That is still true. But the more important prediction, the one I care more about, is the positive version of the same moment.
Within 24 months, the enterprises that built the right governance infrastructure will be running hundreds of AI agents in their core business processes, Finance, HR, Sales, Operations, IT, with the confidence to keep expanding the scope of what those agents are trusted to do. They will be operating what Ryan Knisley, former CISO of Disney and Costco and one of our advisors, described as an identity fabric for agents: a single governance layer that any agent in the enterprise can connect into, with consistent identity, consistent oversight, and consistent accountability.
The organizations that get there will not just be more productive. They will be structurally different from the ones that didn't, with a digital workforce operating at machine speed across their most complex business processes, governed with the same accountability standards they apply to their human workforce.
That is the revolution. Not the technology. The trust infrastructure that makes the technology's full potential accessible.
I’ll leave you with this
We started aizome because we believe enterprise AI agents are going to transform how work gets done, and because I believe the identity and governance infrastructure to make that transformation safe and accountable is the most important problem in enterprise security right now.
Identiverse 2026 confirmed that belief. The community is ready for this conversation. The analyst community is naming the category. The business case is proven. The governance infrastructure is being built.
The question for every enterprise leader is no longer whether to deploy AI agents. It is whether to build the governance infrastructure that lets you trust them with the work that actually matters.
That is the opportunity. And it is bigger than most people in the security industry are willing to say out loud.
Amir Ofek is CEO and Co-Founder of aizome, the leading Enterprise AI Agent Identity Fabric Platform. He presented at the Non-Human Identity & Agentic AI Summit at Identiverse 2026 and participated in the panel "The Next 24 Months: Predictions on Where NHI & Agentic AI Security Goes Next."
