Map / Sandboxing
Sandboxing
Isolated execution environments where AI-generated code and agent actions can run without touching production. Container, microVM, and syscall-level boundaries that contain blast radius when agents go off-script.
Quick answer: The map tracks 25 companies in Sandboxing; 9 list it as their primary category. The most heavily funded are Teleport ($169M), Duality Technologies ($49M) and ARMO ($34.5M). Most are based in USA (12), Israel (4) and Canada (2). 1 has been acquired.
25 companies, A–Z
- ARMOCloud Application Detection & Response (CADR) built on Kubescape (CNCF project, ARMO is the originator); Application Profile DNA (per-agent behavioral baseline of tool usage, network connections, data access — built dur…
- AutonomousEndpoint-native AI agent security platform restoring control over agent execution; three integrated layers — visibility, control, and isolation; instant discovery of every installed AI agent, skill, plugin, and MCP serv…
- BlaxelCloud infrastructure platform purpose-built for AI agents ("AWS for AI Agents"); sandboxed virtual machines that scale to zero after 1s of inactivity and resume in under 25ms with full memory state preserved; agent co-h…
- Canyon RoadExecution-layer security for AI workloads with three products — agentsh (open-source runtime enforcing least privilege at the syscall level for unsupervised headless agents in CI, containers, pipelines, dev environments…
- DaytonaAI agent execution sandbox infrastructure — isolated execution environments for AI-generated code with sub-90ms cold starts (27ms in optimized configurations); SDK + CLI for AI agents to create/use/destroy sandboxes on…
- Duality TechnologiesSecurePlus privacy-preserving data collaboration platform combining homomorphic encryption with data science; secure multiparty data collaboration using TEEs for AI/ML workloads; AI assistant for sensitive data analysis…
- EderaSecure infrastructure for deploying and executing AI agent workloads with strong isolation; Type 1 hypervisor-grade isolation (heritage from Xen / unikernel work) applied to container workloads running AI agents; multi-…
- Galene.AISovereign agentic AI platform deployable on-prem or private cloud; Generative Shield runtime guardrails and cyber security inspection; personal AI assistant + specialized AI agents + customizable AI agents; MCP-based co…
- GuardiAgentSandboxes AI agents and MCP servers in isolated containers with least-privilege, user-consent permission models and auto-generated security manifests.
- I-THAMCybersecurity platform purpose-built for AI, defending models, agents, data, and AI-powered applications with continuous monitoring across the AI stack — from prompt to tensor.
- IsloSecure, long-running sandbox for AI coding agents: each agent gets its own dedicated, persistent cloud environment (a real Linux microVM, not docker-in-docker) that runs real services, survives reconnects, and resumes w…
- JozuJozu Agent Guard — zero-trust AI runtime that executes agents, models, and MCP servers inside secured environments with policy enforcement and guardrails that agents cannot disable; vetting, signing, and governing AI ar…
- KeyboardSecure control plane for AI sitting between AI clients (Copilot, Cursor, Claude, Claude Code, Gemini, ChatGPT, Open WebUI) and company applications; organization-wide visibility and policy enforcement for AI-driven auto…
- LoricaConfidential AI infrastructure layer enabling privacy-preserving model deployment for AIaaS providers and enterprises; hardware-backed security and attestation at every layer; one-click deployment of confidential model…
- Mithril SecurityConfidential AI inference allowing models to run on private data inside hardware-based trusted execution environments.
- NopeIDOn-device security for AI agents running on employee endpoints.
- ObliviousPrivacy-preserving AI agent platform for running LLMs on sensitive data without exposing it to the model provider; confidential computing layer for AI workloads enabling regulated industries to use LLMs without sending…
- OpaqueConfidential AI platform enabling enterprises to run cloud-scale general-purpose AI workloads on encrypted data with hardware-attested execution; Confidential Agents for RAG and agentic workflows; Opaque Studio for acce…
- PipeLabPipelock — open-source agent firewall for MCP and AI agent egress; 11-layer scanner pipeline with 48 DLP credential patterns (Base64/hex/URL/Unicode encoding-aware); 25 prompt-injection detection patterns with 6-pass no…
- RunloopEnterprise-grade infrastructure platform for developing, evaluating, and deploying AI coding agents at scale; Devboxes — secure micro-VM sandboxes with dual-layer isolation (VM + container) on a custom bare-metal hyperv…
- SecretariumKlave for AI confidential computing platform for private, verifiable AI inference on sensitive data; trusted execution environment-based deployment; deterministic and verifiable computation; RAG security; integration wi…
- TeleportEstablished identity-based infrastructure access platform (SSH, RDP, HTTPS, Kubernetes, cloud consoles via built-in proxy) with dedicated Agentic AI Security use case; eliminates shared secrets, standing privileges, and…
- Tenet SecurityRuntime Agent Defense layer for AI agents that monitors, controls, and secures agent reasoning and actions in real time with adaptive guardrails, shifting from passive monitoring to active enforcement.
- TinfoilConfidential AI platform that runs models inside secure hardware enclaves, offering private inference, a private chat and an OpenAI-compatible inference API with zero-trust, zero-access, zero-retention guarantees; colla…
- Valarian TechnologiesACRA AI private AI model hosting platform for on-premise deployment in secure environments; zero-trust architecture; tactical/defense-focused AI workloads; air-gapped and disconnected deployment support; broader Valaria…
Other categories
Observability & Governance · Runtime & Guardrails · AI Identity · MCP & LLM Gateways · AI Red Teaming · AI CSPM · AI Data Governance · Model Security · AI Network Security · AI Endpoint Security · AI Code Security · AI Browser Security · AI SSPM