AI Governance & Security Vendor Wiki

7 min read

Oasis Security

A Non-Human Identity Management (NHIM) platform: discover, posture-check, govern, and remediate the service accounts, tokens, secrets, workloads — and now AI agents — that outnumber human identities in the enterprise.

One-liner — A platform that finds, governs, and cleans up all your non-human identities (service accounts, API keys, tokens, bots, workloads), and has extended that into purpose-built identity governance for autonomous AI agents.

Categories — non-human-identity

What it does

Modern enterprises run far more non-human identities (NHIs) than human ones — service accounts, API keys, OAuth tokens, certificates, secrets, and increasingly AI agents — and most are ungoverned, over-privileged, and unrotated. Oasis builds an inventory of every NHI across cloud, SaaS, and on-prem, classifies them, scores their posture (stale/unused identities, standing secrets, excessive privilege, anomalous behavior), and drives remediation including safe secret rotation. Its own framing is “discover, resolve, automate.”

In late 2025 Oasis launched Agentic Access Management (AAM) (announced 2025-11-19), which it markets as the first identity solution purpose-built to govern AI agents across their full lifecycle. AAM adds three things on top of the NHI base: intent inference (LLM-driven reading of what an agent is trying to do, so policy can judge purpose, not just the raw action), deterministic policy enforcement (a policy engine that approves/denies and escalates to a human when an agent crosses a privilege boundary), and just-in-time session identities (ephemeral, least-privilege credentials minted per session to eliminate standing secrets, with a full audit trail).

Where it sits in the stack

Primary category: non-human-identity (Foundation layer). Oasis governs the identities that machines and agents use, which is upstream plumbing rather than an AI-specific control. By enforcing least privilege and just-in-time, short-lived credentials it limits what a compromised workload or a misbehaving agent can reach and exfiltrate. It does not inspect prompts or model output, so it does not screen untrusted input. Its AAM line reaches into agent-governance territory adjacent to ai-spm and authorization layers (authorization-engine), but its spine remains identity governance, not runtime prompt security.

Deployment & architecture

  • SaaS, agentless. No endpoint agents; Oasis connects via API/read integrations to discover and govern NHIs. This lowers deployment friction versus agent-based approaches.
  • Integrations (per vendor site, marketing): IdPs and clouds — Azure, AWS, GCP, Okta, Ping, Active Directory; secret vaults — HashiCorp Vault, Azure Key Vault, AWS KMS, GCP; PaaS/SaaS — Snowflake, Databricks, GitHub, Salesforce; AI services — OpenAI, AWS Bedrock, Glean.
  • Capabilities map: discovery/inventory, posture management (incl. an AI-SPM-style agent risk view and anomaly detection), lifecycle (provisioning to deprovisioning, secret rotation across vaults), and remediation with time-bound, intent-aware access.
  • AAM issues ephemeral just-in-time session identities and runs an LLM-driven intent-inference plus deterministic policy engine for AI agents.

Positioning & differentiators

Oasis is one of the category-defining NHI startups and is known for an agentless, fast-to-stand-up discovery+remediation platform and an early, aggressive pivot to agent identity (AAM). It competes with a crowded NHI field:

  • token-security and entro-security — NHI discovery/posture and secrets-security peers; Entro leans secrets-centric, Token leans machine-identity-centric.
  • clutch-security — NHI lifecycle/security peer.
  • astrix-security — closest direct rival on NHI + agent identity (note Astrix carries a seed flag of acquisition by Cisco — verify on its own page).
  • aembit — adjacent but architecturally different: a workload IAM / access “broker” that brokers credentials at access time, vs Oasis’s discover/govern/posture model.
  • cyberark — the incumbent identity-security platform (secrets via Conjur, plus NHI and machine identity); the big-vendor consolidation threat to point NHI startups.

Oasis’s differentiator pitch is intent-aware, lifecycle-wide governance (“access control that understands intent, not just static roles”) and being first to market with a dedicated agentic access product. These are vendor claims; independent corroboration of “first” is weak.

Ownership, funding & M&A

Independent, venture-backed. Founded 2022 by Danny Brickman (CEO) and Amit Zimerman (Chief Product Officer), both veterans of Israeli military cyber (reported as Unit 81 / IDF cyber R&D). Funding timeline:

  • Stealth exit 2024-01-31 with $40M total ($5M seed + $35M Series A), Series A led by Sequoia Capital, with Accel, Cyberstarts, Maple Capital, and angels Guy Podjarny (Snyk) and Michael Fey (Island).
  • $35M Series A extension, May 2024 (led by Accel, Cyberstarts, Sequoia per press).
  • $120M Series B, 2026-03-19, led by Craft Ventures, with Cyberstarts, Sequoia, and Accel. Reported total to date ~$195M (one source says $190M — minor unreconciled discrepancy). Company reports new ARR up 5x YoY and a majority-Fortune-500 customer base.

No acquisition found — the seed registry carried no M&A flag, and nothing in research indicates one. Ownership: independent / confidence high (multiple dated primary/press sources on funding; no cap-table doc reviewed).

Soft discrepancy: founding year given as 2022 by Oasis’s own 2025–2026 materials and the Series B coverage; the Jan-2024 TechCrunch piece implied ~2023 (“less than a year after founded”). HQ is now New York, NY per current vendor/PR materials, with Tel Aviv R&D roots (early coverage called it Tel Aviv). Using 2022 / New York. Non-blocking.

CTO / hedge-fund lens

NHI governance is generally Day-2 for a hedge fund: you stand up human IdP/SSO, secrets management, and logging first (Day-1), then tackle the sprawl of service accounts, tokens, and secrets. It becomes more pressing — arguably Day-1.5 — the moment you start deploying AI agents with real access to data and systems, which is exactly the gap Oasis’s AAM targets: ungoverned, over-privileged agent credentials are a direct path to data egress. There is no specific SR 11-7 / model-risk mapping here; this is a security/identity control, not a model-governance tool.

Fit caveats: Oasis sells primarily to Fortune 500 / large enterprise, so a 50-person fund may find it heavier and pricier than needed, and may get partial NHI coverage from an existing cyberark footprint or cloud-native IAM. The agentless SaaS model is the friendliest part of the deployment story. A mid/large asset manager already running many service accounts across Snowflake/Databricks/GitHub and piloting agents is the natural buyer.

Competitors / alternatives

astrix-security, token-security, entro-security, clutch-security, aembit, cyberark, silverfort, natoma

Open questions / to verify

  • Exact cumulative funding ($190M vs ~$195M) and precise structure of the May-2024 $35M extension.
  • Founding year 2022 vs 2023 (use 2022 per vendor; TechCrunch implied later).
  • Real-world maturity/adoption of AAM (launched 2025-11) vs marketing; whether “first” agentic identity product holds up against astrix-security and others.
  • Whether any on-prem/self-hosted option exists or it is SaaS-only.
  • Pricing model and minimum viable deployment for a mid-market shop.

Sources

History

  • [2026-06-28] Stub created from seed registry.
  • [2026-06-28] Researched; established founded 2022, HQ New York NY (Tel Aviv R&D roots), founders Danny Brickman (CEO) & Amit Zimerman (CPO), independent/venture-backed with ~$195M total to a $120M Series B (2026-03-19, Craft Ventures); no M&A. Documented agentless SaaS NHI platform (discover/posture/lifecycle/remediate) and the 2025-11 Agentic Access Management launch (intent inference, deterministic policy, JIT session identities). Positioned vs Astrix/Token/Entro/Clutch/Aembit/CyberArk. Set status: researched, confidence medium, ownership_confidence high. 4 sources cached. Noted soft discrepancies on founding year (2022 vs 2023) and total funding ($190M vs ~$195M).

Graph View

Backlinks