Map / MCP & LLM Gateways
MCP & LLM Gateways
Protocol-layer control plane sitting between agents and the MCP servers, tools, and models they call. Per-request auth, scope enforcement, tool-call inspection, and supply-chain governance for the agent ↔ tool boundary.
Quick answer: The map tracks 115 companies in MCP & LLM Gateways; 36 list it as their primary category. The most heavily funded are C1 ($111M), Cequence Security ($100M) and Neo ($100M). Most are based in USA (66), Israel (24) and India (5). 8 have been acquired.
115 companies, A–Z
- AcuvityComprehensive AI security and governance platform across employees and applications; intent-based access controls; real-time monitoring of prompts, responses, and agent behaviors; visibility into GenAI applications, ser…
- AembitWorkload IAM extended to AI agents; Blended Identity model combining the AI agent's non-human identity with the human operating it in a single policy decision at request time; MCP Authorization Server (OAuth 2.1 authori…
- agen.coBehavior-based unique agent identity, trust levels + risk scores, RBAC + entitlements + feature-level permissions, contextual policies, PII redaction, scoped data access, behavior anomaly detection, delegated permission…
- Agentic Fabriq"Okta for Agents" — identity and permissioning layer for AI agents; per-agent identity with least-privilege access; full audit logs of every action every agent takes; one-click revocation across MCP, OAuth, and agent fr…
- aggreteOpen-source MCP proxy sitting as a governance layer between AI assistants/agents and backend data connectors (Slack, GitHub, Drive, Jira, Salesforce, Notion), enforcing a YAML-defined policy on every tool call.
- AI AramidEnterprise AI security control plane sitting between applications and every LLM (cloud or self-hosted), controlling every request before it leaves the network: a traffic gateway, real-time threat blocking, automated red…
- AI Security GatewayOpenAI-compatible proxy for LLM applications providing DLP, PII redaction, and cost governance; AI gateway and AI firewall functionality; prompt injection detection; shadow AI discovery; works as drop-in replacement for…
- Aim SecurityThree-pillar AI security platform: secure employee use of public AI applications (Microsoft Copilot, ChatGPT, coding agents); secure private AI applications and AI agents via Aim AI Firewall enforcing corporate policy o…
- AIRSecures every layer of the agentic stack — skills, plugins, MCPs, and sub-agents, both external and internally built — so enterprises can adopt AI safely.
- aizomeControl layer between employees building AI agents and the enterprise systems those agents access: automatic discovery of every AI agent (including unsanctioned shadow agents), risk scoring, identity issuance, and permi…
- Akeyless SecurityThree agent-facing products on top of the secrets platform: SecretlessAI brokers agent access through the Akeyless Gateway so agents never hold a secret; Agentic Runtime Authority evaluates each request against identity…
- AptaSentryGoal-based automated red teaming across text/audio/video/image (AptaRed), agentic workflow stress-testing for prompt injection/privilege escalation/data exfil (AgentRed), model risk evaluation with findings fed back int…
- AptibleAI control layer with an AI Gateway and MCP Gateway that lets teams use any model, connect any tool and deploy any agent under enforced access control.
- Arcade.devMCP runtime for production AI agents; Contextual Access (per-action authorization based on user identity, tool intent, data sensitivity — featured by Snyk as the reference pattern for agent guardrails); secure agent aut…
- ArchestraOpen-source MCP orchestrator for enterprise AI agent deployment; helps non-technical teams securely connect AI agents to enterprise data via MCP; dynamic permission controls; security + orchestration layers added on top…
- Arrakis SecurityAutonomous Workforce Governance Platform delivering total visibility, policy enforcement, and real-time remediation over autonomous AI agents operating within enterprise infrastructure, spanning coding agents, assistant…
- AuthPlaneSelf-hosted, spec-compliant OAuth 2.1 authorization server for MCP servers, shipped as a single Go binary that deploys alongside an MCP server and stands up a fully compliant authorization layer in under 10 minutes.
- 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…
- BackslashVibe coding security platform; end-to-end protection for AI-native software development stack covering IDEs, AI coding agents (Cursor, Claude Code, Windsurf, Gemini CLI, GitHub Copilot), MCP servers, prompt workflows an…
- Barndoor AIControl plane for orchestrating, securing, and governing AI agent workflows across an organization.
- BeyondGuardEnterprise AI security control plane with dedicated guards across six layers: RAG/data ingestion, LLM inference and model supply chain, prompt orchestration, agent actions, MCP/tool integrations and response delivery; A…
- BlueRockAgentic observability across the full Agentic Action Path (model decision → tool call → MCP server → data access → outcome); MCP Trust Registry with verified server ownership/capabilities; Trust Context Engine carrying…
- BritiveBritive ARC (Agentic Runtime Control) applies a verify-assess-authorize-observe-revoke loop to every privileged AI agent action: agent discovery in cloud platforms that onboards each agent as a managed identity with a n…
- BurgusSecurity and audit gateway sitting between AI applications and their existing LLM or MCP providers, inspecting model and tool traffic in real time and combining per-request policy checks with behavioral learning across…
- C1AI-native unified identity governance for human, non-human, and AI agent identities (rebranded from ConductorOne 2026); AI Access Management control plane for AI tools, agents, and MCP connections; Non-Human Identity Go…
- Cequence SecurityEstablished API security + bot management vendor extended to agentic AI; AI Gateway with 140+ verified enterprise application integrations; Agent Personas launched April 2026 — industry-first infrastructure-level privil…
- CloudMatosAegis Gateway — runtime security gateway specifically designed for multi-agent AI systems; policy enforcement across agent-to-agent communication; runtime guardrails for complex agent orchestrations; cloud security post…
- CodeIntegrityRuntime control plane that sits between AI agents and enterprise systems, mediating every agent action before execution; turns agent instructions into sandboxed code so each action is explicit, inspectable, and repeatab…
- CredalBuilds, governs, and deploys enterprise AI agents and MCP servers powered by company data under centralized security and governance controls.
- CyBox SecurityAI security platform spanning Visibility, Governance, and Enforcement: discovers which AI tools, models, agents, Skills, and MCP servers are in use, who is using them, and what they can access, then turns policy into re…
- Dash SecurityAgentic AI security platform that discovers, governs, and protects every agent, MCP server, and skill in real time — giving enterprises the clarity, control, and confidence to let agents run without slowing innovation d…
- DatawizaAgent Gateway that sits between AI agents and the systems they reach — enterprise APIs, MCP servers, SaaS apps, DevOps and observability platforms, databases and LLM providers — enforcing identity-aware access.
- Daxa.aiData-first AI governance and security platform; Pebblo AI data platform embedding access control, security, and governance into AI apps with developer-friendly hooks; Proxima secure AI knowledge retriever for role-appro…
- DeepTrailAdds auth and permissions to MCP servers and AI agents; identity/credential/access management for AI-native apps.
- DescopeAgentic Identity Hub 2.0 — agents as first-class identities with dedicated OAuth client IDs and tool-level scopes; comprehensive MCP auth (OAuth 2.1, DCR, CIMD, per-tool scopes, multi-tenant); credential vault for OAuth…
- EnforceAuthCentralized authorization control plane (OPA/Rego/Cedar) enforcing fine-grained policy across apps, data, cloud, and AI agents, controlling agent-to-data and agent-to-agent actions in real time.
- FencioZero-trust firewall for AI agents: learns an agent's behavioral boundaries during pre-production evaluation (Shark), then enforces those constraints deterministically at runtime (Prism), outside the LLM itself.
- First Recon AIEnd-to-end enterprise AI control and security platform built across three layers: on-device semantic inspection that evaluates sensitive data, context, and intent before content reaches any AI provider and supports loca…
- ForceAI SecuritySingle platform (Pinesmith) to secure AI activity, prompts, agents, MCP servers, and workflows from design-time guardrails through runtime, plus governed AI agents for security operations.
- Full Court DefensePer-machine agent and CLI that checks every action an AI coding tool or agent takes before it runs: shell commands, file edits, MCP tool calls and API calls.
- Future AGIEnd-to-end AI agent engineering platform covering the full lifecycle: simulate (multi-turn scenarios, synthetic data generation, persona testing), evaluate (factuality, groundedness, hallucination, PII, toxicity, tool s…
- GolfAgentic AI Security and Governance control plane at the MCP layer (not the LLM layer) — explicit thesis that "AI gateways sit between your app and the LLM — useless for third-party agents like Claude, Copilot, and ChatG…
- Gopher SecurityAI-powered, post-quantum zero-trust networking platform plus on-demand MCP servers/gateways connecting agentic workflows to enterprise stacks.
- GuardiAgentSandboxes AI agents and MCP servers in isolated containers with least-privilege, user-consent permission models and auto-generated security manifests.
- Helmet SecurityMCP server discovery, traffic monitoring, registry of verified MCP servers/skills, drift detection, supply-chain analysis, secret scanning.
- HighflameUnified AI Security Platform spanning build-time to run-time for GenAI + Agents + MCPs; ZeroID for verifiable cryptographic identity for AI agents (scoped delegation, instant revocation); real-time protection against em…
- HOL GuardOpen-source (Apache-2.0) local security layer for AI coding agents.
- IndyKiteContext-aware authorization platform for autonomous AI agents — AgentControl product governs agents with identity context, policy controls, and audit trails; identity knowledge graph capturing relationships between user…
- Intellect MachinesScouter product: on-demand prompt scanning, jailbreak/prompt-extraction detection, system-prompt exposure detection, real-time risk scoring; 250+ attack probes across 20 categories; multi-agent red-team testing; remedia…
- Invariant LabsSecurity and reliability platform for AI agents and MCP servers; Guardrails contextual security layer enforcing deterministic rules on data flow, tool calls, and content; Explorer runtime observability dashboard inspect…
- KeelAI control plane that issues signed "AI Permits" — pre-execution decision records (allow/deny/throttle/review) — for consequential AI agent actions, LLM provider calls, and MCP tool calls, enforcing policy and budget li…
- 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…
- KIPIOAI Operations & Security platform purpose-built for MSPs, giving them full visibility and real-time protection across all AI surfaces used by their clients, delivered as a repeatable managed service.
- LakeraAI-native security platform for agentic AI applications; flagship products Lakera Red (continuous AI red teaming) and Lakera Guard (real-time runtime enforcement); detection rates above 98% with sub-50ms latency and fal…
- LangProtectThree products.
- Lasso SecurityAgent discovery via platform/CI integration, AI security posture analysis, adversarial testing / purple teaming with 3,000+ attack library across OWASP Top 10, multi-turn agentic attacks, runtime guardrails, MITRE/OWASP…
- LevoUnified API + AI security platform: runtime discovery of LLMs, agents, MCP servers and RAG pipelines, AI-SPM, AI red teaming, AI gateway/firewall, MCP discovery and security testing.
- Linx SecurityIdentity security and governance platform (IGA + ISPM) built on an identity graph spanning humans, non-human identities and AI agents: access requests and reviews, joiner-mover-leaver automation, posture findings (dorma…
- LoopersBare-metal, zero-delay circuit breaker for AI API billing that sits as a transparent reverse proxy between applications and LLM providers, enforcing spend limits before requests reach the model.
- LumeusZero Trust gateway for GenAI/LLM and Cloud Access with MCP protocol support; AI-native security for modern development teams protecting AI coding workflows, MCP protocols, and infrastructure access; ephemeral credential…
- LumosAutonomous identity governance (IGA) platform that governs access for human, non-human and AI agent identities from one live map of identities and fine-grained app permissions.
- Lunar.devMCPX — a gateway for securing, governing, and auditing AI agent access to MCP servers; RBAC and SSO enforcement at the MCP layer; observability into MCP server usage; AI gateway capabilities for routing and securing age…
- MintiveEnterprise MCP gateway + LLM proxy monitoring every tool call, bash command, and file operation from coding agents; role-based MCP endpoints; one-click hosting of STDIO-based MCP servers as managed enterprise services;…
- NatomaEnterprise authentication, fine-grained authorization, and governance for AI agents via a hosted MCP platform.
- NeoPrevents agentic threats across every device, browser, identity, and traditional application in the enterprise, giving SecOps real-time visibility and control over all human and non-human activity.
- NeuralTrustComprehensive AI security platform spanning gateway, runtime, and testing: TrustGate (AI gateway with authorization, rate limiting, content filtering across LLM providers); TrustLens (observability and runtime monitorin…
- Nexos.aiUnified AI platform combining workspace interface and AI gateway; multi-LLM access with secure proxying; agent governance; DLP for prompt content; cost management and observability across providers; positioned as a neut…
- NyuwayAgentic AI security platform that discovers, governs, and protects AI agents and MCP connections across the enterprise, combining an MCP security gateway with per-agent identity and AI red-teaming.
- Obot AIComplete open-source MCP platform spanning hosting + registry + gateway + chat client; Obot MCP Gateway as centralized control plane; administrative control plane for IT to onboard, manage, and update MCP servers via mo…
- OpenShieldTransparent proxy firewall providing rate limiting, content/keyword filtering, tokenizer/cost tracking, and model pipelining for AI models.
- Operant AI3D Runtime Defense (Discover/Detect/Defend), Endpoint Protector, Agent Protector, AI Gatekeeper, MCP Gateway, API protection, Woodpecker OSS red teaming.
- OryoEnterprise security control plane for AI agents; discovers and secures every AI agent + tool + MCP + data flow + identity + model + prompt in a single map; connects to SaaS, cloud AI agent platforms, identity systems, a…
- Pallas SecurityApp-aware, object-level access control that sets policies per file, folder, or label.
- Permit.ioEnterprise-grade AI agent authorization platform; OPAL (Open Policy Administration Layer) — open-source policy administration extended for AI agent contexts; fine-grained per-action authorization decisions for agent too…
- 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…
- PlainIDPolicy-Based Access Control (PBAC) platform — Discover, Manage, and Authorize modules — repositioned as the runtime authorization layer for agentic AI.
- Pluto SecurityAI workspace security for the AI tools employees use, the components plugged into them, and what employees build with them.
- PointGuard AIFull-stack AI security and governance platform (formerly AppSOC, rebranded April 2025); AI Discovery and Inventory across models, agents, MCP servers, and integrations; AI Security Posture Management (AI-SPM); AI Securi…
- PolaxisAI agent security and governance platform that intercepts every tool call through a 7-layer firewall — regex scan, risk scorer, LLM gate, behavioral baseline, session graph, threat intel, and policy engine; blocks promp…
- PomeriumOpen-source zero-trust identity-aware proxy with explicit MCP authorization support; per-request authorization with JWT identity and full audit logging; policy-based authorization preventing prompt injection attacks; cl…
- PortkeyUnified control plane for production AI; open-source AI Gateway processing 1T+ tokens and 120M+ AI requests daily across 24,000+ organizations; MCP Gateway for governing AI agents across enterprise tools; observability,…
- Principled IntelligenceAI governance and evaluation suite for agentic systems: Spectral (autonomous eval running thousands of tests to surface failure modes pre-deployment), Orbitals (real-time, sub-second guardrails blocking policy violation…
- PrizmaA self-hosted AI gateway and firewall that inspects requests from employees, coding agents and AI apps to cloud and local models.
- Prompt SecurityFull-stack GenAI and agentic AI security platform; runtime protection at the point of interaction (browser, desktop, API); real-time visibility into AI tool usage across the enterprise; protection for employee use of pu…
- PromptGuardLLM firewall and guardrail API that scans prompts and responses for prompt injection (with normalization of homoglyphs, zero-width characters and encoded payloads), jailbreaks (LLM classifier, including multi-turn escal…
- PromptHaloModel-agnostic platform that combines automated AI red-teaming (static probes, a prompt-injection library and PAIR adaptive attacks) with runtime enforcement over what AI agents are allowed to do.
- PVMLSecure AI-ready virtual databases — enables CIOs / IT teams to securely operationalize GenAI on existing infrastructure without moving or duplicating data; unlimited virtual databases with built-in security and AI readi…
- PyntChain-aware MCP security — agent-based security solution understanding the full MCP chain (multiple servers, tools, and downstream calls) rather than per-call inspection; API security heritage extended to MCP protocols;…
- QueryPieAccess control platform extended to AI, exposing its policy engine as an MCP server so agents operate inside the same data and infrastructure access rules as human users.
- RequestyUnified LLM gateway and OpenAI-compatible API for 400+ AI models (Claude, GPT, Gemini, DeepSeek, Llama, Mistral) — drop-in replacement for OpenAI SDK with a single base URL change; intelligent routing automatically dete…
- RevaContinuous adaptive authorization platform for humans and AI agents, positioned as an Authorization Management Platform (AMP) rather than an IdP.
- Rogue SecurityAI agent runtime security across four pillars: AI-SPM (shadow AI discovery, agent inventory/classification), AIDR (detection & response), AI AppSec (build-time/CI-CD testing with the Rogue OSS red-team engine), and Acce…
- RunlayerPrivate MCP/skill/agent registry, one-click MCP installs, custom MCP hosting, subagent orchestration, identity-bound permissions, MCP attack detectors.
- ScalekitAuth stack purpose-built for AI agents: OAuth 2.1 authorization server for MCP servers, delegated/ephemeral agent identity, per-tenant encrypted token vaulting, and enterprise SSO/SCIM, so agents act as specific scoped…
- SenthexEU runtime layer sitting between AI agents and model providers (OpenAI, Anthropic, Mistral, Azure OpenAI), hosted in Germany for data sovereignty.
- SlashIDAI Identity Governance launched May 2026 — first access-graph-native solution governing OAuth-connected AI apps + agents + MCP servers; Access Graph models OAuth scopes as first-class edges; continuous discovery of OAut…
- SlashLLMISP-style "Integrated Service Provider" for AI security — platform (SlashStack) + embedded operations team + governance program in one offering; unified API gateway with authentication, rate limiting, and policy enforce…
- Sonoma SecurityMCP governance platform for securing and controlling enterprise AI agents; supply-chain security for MCP servers used by AI agents; permissions enforcement on agent ↔ MCP interactions; prompt injection protection at the…
- Speakeasy"AI control plane" for enterprise agent fleets, centered on an MCP Gateway that fronts every agent-to-MCP-server call through one URL.
- StacklokEnterprise MCP platform managing authentication, authorization, and policy controls for AI agent access to tools and systems; runs in customer's VPC (not SaaS) for data residency-sensitive deployments; ToolHive open-sou…
- StraikerDiscover AI (AI-SPM), Ascend AI (autonomous red teamer), Defend AI (runtime guardrails); MCP security.
- Sun SecurityAI-native agent security platform ("Own Your Agents") giving enterprises visibility, governance, and runtime control over AI agents across the full lifecycle: discovery, risk assessment, posture management, runtime prot…
- Tego AIAgent-native security platform built on three pillars: discover, monitor, and control AI agents across the enterprise.
- TetrateTetrate Agent Router Enterprise (TARE) — GenAI runtime visibility and governance platform built on Envoy service mesh heritage; LLM traffic management routing requests across model providers; AI data gateway capabilitie…
- TraceforceAutomatically discovers every AI tool, MCP, and skill across endpoints and assesses risk from actual usage patterns.
- Ultra SecurityRuntime security platform for AI agents.
- UnboundGenAI security platform and AI Gateway that discovers and governs employee GenAI usage and intelligently reroutes sensitive prompts to secure self-hosted open-source models; now an Agent Access Security Broker (AASB) th…
- VerosekAI + MCP gateway that sits as one layer between apps and every AI agent, acting as a drop-in replacement for the OpenAI, Anthropic, and Gemini SDKs.
- VorlonAgentic Ecosystem Security Platform; AI Agent Flight Recorder — captures continuous cross-app audit trail of every agent action (every identity, SaaS app, API endpoint, data classification, downstream system touched) bu…
- VulnifyRuntime authorization layer for AI agents: before a tool call executes, the agent asks Vulnify for a decision and gets ALLOW, REVIEW or BLOCK with a 0-100 risk score and reasons.
- WallarmAPI-first security platform with dedicated "Protect Agentic AI" product line; extends established API security (WAAP, API discovery, schema enforcement, anomaly detection) to AI agents and AI-enabled APIs; agent-aware t…
- WardlineInline gateway in front of LLM traffic that returns one of five decisions per request (ALLOW, MASK, REWRITE, REQUIRE_APPROVAL, DENY) with reason codes.
- WaxellThree products — Connect (MCP coordination: policy checks, PII scanning, audit trails per tool call, rug-pull detection, human-in-the-loop approvals); Observe (auto-instrumented tracing across LLM/tool calls, 50+ policy…
- Willow"Basecamp" for bring-your-own AI agents — every agent gets an identity, scoped permissions, and full audit trail; 1000+ connectors; API-to-MCP conversion; reusable pre-permissioned skills bundling multiple tools; shadow…
- Xage SecurityUnified Zero Trust for AI layered on the Xage Fabric.
Other categories
Observability & Governance · Runtime & Guardrails · AI Identity · AI Red Teaming · AI CSPM · AI Data Governance · Model Security · AI Network Security · AI Endpoint Security · AI Code Security · AI Browser Security · AI SSPM · Sandboxing