Map / MCP & LLM Gateways / Speakeasy

Speakeasy

USA · Founded 2022 · Raised $26M

Quick answer: Speakeasy is an AI security company based in the USA and founded in 2022, working in MCP & LLM Gateways, Observability & Governance, AI Identity and 1 more category. It has raised $26M from investors including FPV Ventures, GV and Quiet Capital.

MCP & LLM GatewaysObservability & GovernanceAI IdentityRuntime & Guardrails

Capabilities

"AI control plane" for enterprise agent fleets, centered on an MCP Gateway that fronts every agent-to-MCP-server call through one URL. Central catalog of approved agents, MCP servers, and skills; shadow-AI discovery; SSO-backed agent identity; runtime guardrails that block prompt injection, PII, and credential leakage; full audit trail. Three paths to a production MCP server — generate from an OpenAPI spec, import an existing TypeScript server, or deploy pre-built integrations (Stripe, GitHub, Slack) — plus AI Cost Control, which consolidates spend and token usage across Claude Code, Cursor, and Codex per employee, team, model, and session.

Sensors / integration

Gateway-side interception of every MCP session and tool call, with distributed tracing and per-tool latency, error-rate, and usage telemetry. Client-side gateway hooks intercept MCP traffic at the agent endpoint (Claude Web/Desktop, Cursor, Windsurf, ChatGPT, Copilot, internal agents) to surface shadow MCP servers connecting outside the gateway. A device-resident agent collects AI session telemetry normalized to OpenTelemetry Semantic Conventions, tying activity to work identities even when employees use personal accounts.

Protections

OAuth 2.1 with DCR and PKCE normalizing heterogeneous server auth behind one SSO connection (Okta, Entra ID, Auth0, SAML/OIDC); RBAC scoped at server, toolset, and individual-tool granularity, so each team sees only its own catalog; per-agent identities replacing shared service accounts; runtime guardrails that block PII and prompt injection before a call reaches its target; unified audit trail for investigation and EU AI Act / ISO 42001 / NIST AI RMF evidence.

Differentiator

Came to agent security from the API-tooling side rather than the security side: founded 2022 by Sagar Batchu (CEO) and Simon Yu (CPO), the company built OpenAPI-native SDK, CLI, and Terraform-provider generation — still sold as "SDKs & CLIs" — then repointed that spec-to-server pipeline at MCP, launching the MCP Gateway in March 2026. The result governs the agent↔tool boundary as an API problem: the same machinery that generates a server also enforces auth, scope, and guardrails on it, so governance rides on the deployment path instead of bolting on beside it.

Team & funding

Founders
Sagar Batchu, Simon Yu
Total raised
$26M
Investors
FPV Ventures, GV, Quiet Capital

Other MCP & LLM Gateways companies

Acuvity · Aembit · agen.co · Agentic Fabriq · aggrete · AI Aramid · AI Security Gateway · Aim Security · AIR · aizome · Akeyless Security · AptaSentry · See all →

Last updated · Get the “Listed on” badge