Zenity
Type: company Sector: AI-agent security & governance (agent-native platform; adjacent to the AI-governance watchlist rather than on it) First seen: 2026-06-29 (least-agency research ingest) Last updated: 2026-08-14
Created 2026-08-14 from [S-2026-08-03-zenity-series-c] and [S-2026-08-06-zenity-ai-total] (page-creation trigger: third source on the company).
Snapshot
Zenity is an AI-agent security and governance platform (founded by Ben Kliger and Michael Bargury; ~230 staff; R&D Tel Aviv, GTM New York) positioned as governing agents “everywhere” — SaaS agents, coding/personal agents, and custom agentic systems on cloud platforms [S-2026-08-03-zenity-series-c]. It matters to this wiki for three reasons: it is the most heavily capitalised pure-play in the agent-control layer this vault tracks ($125M Series C, Aug 2026); Gartner named it “the company to beat” in AI agent governance (April 2026, vendor-quoted); and its Labs arm keeps producing the incident/threat evidence (AgentFlayer, malicious-skills research) that gives agentic-governance assurance questions concrete substance [S-2026-08-03-zenity-series-c][S-2026-08-06-zenity-ai-total].
Positions / Claims they advance
- “Least agency” over least privilege: agents can act outside their intended purpose while inside their permission set, so permissions must be supplemented by behavioural authorization — decision budgets, runtime scoping [S-2026-06-08-zenity-least-agency]. (Vendor research, secondary-sourced.)
- Pre-execution, intent-aware enforcement: the platform “deterministically allows, modifies or blocks an action before it occurs by understanding an agent’s intent” — Zenity’s answer to this vault’s “blocked vs logged” test, though no evidence-record format or regulatory mapping is disclosed [S-2026-08-03-zenity-series-c]. (Vendor-asserted, unverified.)
- The agent supply chain is a distinct attack surface: skills, tools, MCP servers, packages and web content can all manipulate agent behaviour; static review structurally misses runtime-emergent maliciousness, so components should be “detonated” in a sandbox and judged on observed behaviour (AI Total / Agent Detonation Chamber, free service) [S-2026-08-06-zenity-ai-total]. (Vendor research-marketing pairing.)
- Regulated-industries positioning without named FS references: customer base claimed as majority Fortune 500/Global 2000 “across financial services, healthcare, pharmaceuticals…”; the only named customer is SoftBank Corp (telecom, Japan) [S-2026-08-03-zenity-series-c] ⚠️ vendor claim.
Relationships
- relates-to → AI Governance Platforms — supplies the behavioural-authorization concept and (now) the largest funding event in the security-capitalised agent-control layer [S-2026-08-03-zenity-series-c].
- relates-to → Model Risk Management and Agentic AI — its enforcement and supply-chain claims bear on agentic model-risk control design [inference].
- competes-with → the agent-control cohort on the topic page (Vorlon, Hush Security, Neo, Onyx, HiddenLayer, Credo AI Agent Governor) — overlapping enforcement/inventory loci [inference; no source names the competitive set].
Tracked changes
- 2026-06-08 — “Least Privilege Fails for AI Agents / Least Agency” research published [S-2026-06-08-zenity-least-agency].
- 2026-08-03 — $125M Series C led by Norwest; SoftBank Vision Fund 2, Hitachi Ventures, LG Technology Ventures join; Gartner April 2026 “company to beat” report quoted; revenue claimed tripling annually [S-2026-08-03-zenity-series-c].
- 2026-08-06 — Black Hat USA: malicious AI-agent skills research (250k-install undetected skill; >30% of dangerous skills abusing coding agents as droppers) and free AI Total dynamic-analysis service launched [S-2026-08-06-zenity-ai-total].
Open Questions
- What evidence record does Zenity’s allow/modify/block decision leave, and would it satisfy EU AI Act Art. 12 / SS1/23-grade record-keeping? (Undisclosed.)
- Does the announced European expansion include EU data-residency / sovereign deployment options? (Unstated.)
- Is there any named EU/UK regulated-FS production deployment? (None named in any source to date.)
Sources
- [S-2026-06-08-zenity-least-agency] → S-2026-06-08-zenity-least-agency
- [S-2026-08-03-zenity-series-c] → S-2026-08-03-zenity-series-c
- [S-2026-08-06-zenity-ai-total] → S-2026-08-06-zenity-ai-total