Hush Security raised $30M in a Series A led by Akamai Technologies on July 28, 2026, to build governance infrastructure for the enterprise AI agent workforce: central registries, scoped just-in-time permissions, and kill switches that apply to every autonomous agent action. The company’s timing is deliberate. Gartner projects the average Fortune 500 will operate more than 150,000 AI agents by 2028, up from fewer than 15 today, and Omdia research finds that 96% of organizations are currently running AI agents on governance models not designed for autonomous software. The governance infrastructure to match that coming scale does not yet exist at most companies.
The Problem: Agents Inherit the Wrong Trust Model
When enterprises deploy AI agents, they typically hand them the same credentials already in use: API keys, service account tokens, standing database access. Those credentials were designed for applications that do exactly what they are told, on demand, and nothing more. AI agents are different. They operate autonomously, initiate actions without a human approval in the loop, and can traverse sensitive systems across dozens of integrations in a single workflow.
Standing credentials handed to an autonomous agent create what Micha Rave, Hush’s CEO, describes as the core governance failure: “AI agents need strict identity, not just API keys. We solved that for non-human identities, and now we’re extending governance to AI agents.”
The problem compounds at scale. Most enterprises do not have a complete inventory of the AI agents running across their organization. Shadow agents, tools built by individual teams without IT oversight, consume the same production credentials as official deployments. There is no central registry, no uniform audit trail, and no mechanism to shut down a specific agent action without revoking broader access.
What Hush Builds: Three Pillars
Hush Security’s platform rests on three connected capabilities, described in the company’s official announcement and reported by SecurityWeek on July 28, 2026.
Discover. The platform identifies every desktop assistant, enterprise AI agent, and custom-built agent across an organization, including unmanaged shadow deployments. It maps the MCPs (Model Context Protocol integrations), tools, and resources each agent can reach and actually uses, producing a live inventory that most enterprises currently lack entirely.
Control. Every agent is enrolled in a central registry and stripped of standing credentials. At runtime, Hush grants scoped, temporary access matched to the specific action the agent needs to perform, and nothing broader. When that action completes, the permission expires. A centralized kill switch can terminate any specific agent action immediately, without revoking broader infrastructure access.
Audit. Every agent interaction produces a complete, immutable log: what the agent attempted, what permissions it was granted, what systems it touched, and what it returned. That log supports security investigations, compliance attestation, and governance reporting without requiring custom instrumentation.
| Governance Challenge | Current Reality | Hush Approach |
|---|---|---|
| Agent credentials | Static API keys with standing access | Scoped JIT permissions, expiring per action |
| Agent visibility | Unknown shadow agents, no inventory | Central registry, all agents enrolled |
| Risk control | Manual, inconsistent revocation | Centralized kill switch per agent action |
| Compliance evidence | No unified audit trail | Complete log per agent interaction |
| Scale readiness | Built for tens of agents | Designed for 150,000+ agents per org |
Who Backed Them and Why
The investor lineup is a meaningful signal. Akamai Technologies, the lead, is one of the oldest and largest infrastructure-grade cybersecurity companies. Its decision to anchor a Series A in an AI agent governance startup reflects its strategic read that identity is the chokepoint in the agentic enterprise.
Ramanath Iyer, Chief Strategist at Akamai, stated in the announcement: “Every major shift in internet usage and traffic has forced a rethink of how enterprises are secured. AI agents are driving the next transformation, and identity is the piece most companies haven’t solved yet. That’s the layer Hush is building.”
Battery Ventures and YL Ventures, both returning from the $11M seed round, doubled their positions. Yoav Leitersdorf of YL Ventures framed the thesis explicitly: “Multiple Fortune 500 companies trusting Hush as they prioritize governance of the AI workforce. Akamai joining this round is a strong signal that the identity gateway for AI agents is quickly becoming core infrastructure.”
Barak Schoster of Battery Ventures added: “Companies are handing autonomous software real authority over real systems, often through the same standing credentials Hush was built to eliminate.”
Total capital raised is now $41M across both rounds.
The Founding Team: Non-Human Identity Experts
Hush was founded in 2024 by Micha Rave (CEO), Shmulik Ladkani (CTO), Alon Horowitz (VP R&D), and Chen Nisnkorn (Chief Customer Officer). All four worked together previously at Meta Networks, which built a zero trust network access platform acquired by Proofpoint in 2019 for $120M.
That background is directly relevant. Meta Networks solved the problem of giving humans the right access at the right time, scoped to a session, with full auditability. Hush is applying the same zero trust principles to non-human identities: secrets, service accounts, and now AI agents. The founders spent years building infrastructure-grade identity controls before AI agents existed as a category. Their architectural instinct predates the current deployment wave.
The company operates with 31 employees across Israel and the United States, with the Series A funding earmarked for expanding both engineering and U.S. sales capacity.
The Anchor Customer: Kyndryl
The most significant validation in the announcement is Kyndryl, the world’s largest IT infrastructure services provider, spun out of IBM in 2021. Kyndryl has deployed Hush internally at global scale and is now reselling the platform to its own enterprise clients.
Adeel Saeed, SVP and CTO Global Cyber Resiliency at Kyndryl, commented: “Hush protects both the autonomous agents and the infrastructure they inhabit, allowing enterprises to deploy AI without sacrificing speed.” The fact that Kyndryl is reselling, not just using, puts Hush’s governance layer inside the procurement and deployment workflows of a company that manages critical enterprise infrastructure for clients globally.
Why This Matters for Enterprise AI Teams
The Hush funding arrives at an inflection point. Enterprises are deploying AI agents rapidly: in sales automation, customer support, data pipelines, code review, and contract intelligence. The infrastructure teams responsible for securing those deployments are running behind the deployment pace.
The failure mode is predictable. Standing credentials granted to a customer-facing support agent can be reused across unrelated systems if the agent is compromised or behaves unexpectedly. A shadow automation agent built by a marketing team may have access to production databases its builders never intended it to reach. Without a central registry, there is no way to know these agents exist, let alone govern them.
The JIT permission model Hush applies is not new as a concept. It is the standard in modern human identity management, where access is granted for a session, scoped to what is actually needed, and logged completely. Applying that standard to AI agents closes a gap that most enterprise security and compliance teams have flagged but few have a clear answer for.
For organizations building on OpenAI’s long-horizon agent infrastructure or deploying autonomous workflows at the scale that platforms like Neo Security addresses with real-time agent control layers, a governance layer like Hush answers a complementary question: not just whether agents stay in bounds during a task, but whether the permissions they hold in the first place are scoped correctly before any task begins.
The architecture question for enterprise AI leaders is now twofold: what agents are running, and what access do they actually have? Hush is building the infrastructure to answer both.
For teams evaluating how to build governance into their AI agent stack before the agent footprint grows beyond what any team can manually manage, Enera works with enterprise clients on exactly this layer.
Sources: SecurityWeek (Ionut Arghire, July 28, 2026), Hush Security Official Announcement via PRNewswire, Calcalist Tech (Meir Orbach), Gartner AI Agent Forecast 2026, Omdia Enterprise AI Governance Research 2026