Company profile
Capsule Security
Runtime AI agent security — discovers agents, monitors behavior, and blocks unauthorized actions before execution (AI circuit breaker).
- Status
- PrivateResearch map: Seed / private · Tel Aviv · $7M (Apr 2026)
- Valuation
- Undisclosed
- Latest funding
- Undisclosed
- HQ
- Boston + Tel Aviv / US-Israel (Startup Radar)
- Last updated
Categories
Funding & valuation
Funding: Undisclosed — no round recorded in the CyberMerge Funding Tracker.
Research map notes
- Research map · AI Agent Security · Undisclosed $7M seed (Lama Partners) — valuation Undisclosed Source (recorded Oct 1, 2026)
Latest signals
- Startup Radar · AI Security · snapshot Sep 29, 2026 2 Sep 2026: launched Nemotron-tuned AI circuit breaker (SiliconANGLE/SecurityWeek); company claims ~96.9% detection / ~71ms decisions vs ~86% strongest third-party evaluated. Prior: ~Apr 2026 stealth exit with $7M Seed (Lama Partners, Forgepoint Capital International). Source
Competitors
Same-category peers from CyberMerge Research maps, Startup Radar and What's The Use tags.
- RecoPrivateIdentity/SaaS graph for apps, AI agents, and NHIs: discovers agents, maps reach to…
- IslandPrivateEnterprise browser plus AI Protect: prompts, tool calls, Island MCP Gateway — agentic control plane for…
- Eve SecurityPrivateAgentic AI runtime security (EveGuard): Agent-in-the-Loop observability, intent/data policy enforcement, and…
- HiddenLayerPrivateModel and agent discovery, AI supply-chain scanning, attack simulation, runtime firewall, AIBOM.
- KontextPrivateRuntime authorization for AI agents: sits between the agent and target systems, checks each requested action…
- OuterlimitPrivateDecentralized security and authorization layer for agentic AI: extends Zero Trust to the agent action layer…
- Token SecurityPrivateAgentless machine and AI-agent identity governance from legacy apps through cloud and SaaS.
- Noma SecurityPrivateFull-lifecycle agentic AI security: discovery/AI-BOM, red team, runtime, MCP and agent access control.
- Pillar SecurityPrivateFull-lifecycle agentic AI security: discovery, agentic red team, runtime, MCP/tool security.
- WitnessAIPrivateNetwork-layer GenAI and agent governance: shadow AI, red team, AI firewall.
- AIR SecurityPrivateInline firewall for AI agents: discover agents, continuously vet skills/plugins/MCP/add-ons, and block…
- Onyx SecurityPrivateSecure AI control plane: discover agents, set permissions, block or redirect risky actions in real time.
Figures appear only as recorded in CyberMerge data, with source and date. “Undisclosed” means no published figure in our data. Editorial, not investment advice. Last updated . Spot an error? Tell the desk.