Teleport is a critical infrastructure player for the AI agent ecosystem, specifically addressing the "authorization gap" that occurs when agents need to interact with production databases, servers, or APIs. Through their Beams product and LLM Proxy, they provide a mechanism to assign cryptographic identities to autonomous agents, ensuring these non-human actors operate under strictly scoped, just-in-time permissions rather than using persistent, high-privilege API keys.
They are active at the security and connectivity layer of the agent stack. By supporting the Model Context Protocol (MCP), Teleport enables developers to build agentic workflows that are production-ready from a security standpoint. For teams deploying agents that can modify live systems, Teleport provides the necessary governance and real-time control to prevent unauthorized actions, making them a foundational component for enterprise agent adoption.
Teleport began in 2015 as Gravitational, founded by the same team that created Mailgun: Ev Kontsevoy, Aleksandr Klizhentas, and Taylor Wakefield. Their core thesis was that infrastructure access was fundamentally broken. Most organizations relied on static credentials—SSH keys, database passwords, and API keys—that were difficult to rotate and easy to leak. Teleport was built to replace these static secrets with a unified identity layer that issues short-lived, hardware-backed certificates for every session.
The company is headquartered in Oakland and has scaled significantly since its Series C funding in 2022. It operates an open-source core with more than 20,000 GitHub stars, allowing small teams to use the product for free while charging larger enterprises for governance, compliance, and advanced security features. Customers include Nasdaq, DoorDash, and Snowflake, who use the platform to manage access across multi-cloud and hybrid environments.
In 2026, Teleport introduced "Beams," a product specifically designed for the AI agent ecosystem. As developers began building agents capable of executing code, querying databases, and managing Kubernetes clusters, the security requirements shifted. An agent acting on behalf of a human should not inherit the human’s full permissions indefinitely. Teleport Beams introduces the concept of Delegated Agentic Identity, which allows agents to prove their identity and obtain scoped, ephemeral privileges to production resources.
This system uses a dedicated LLM Proxy to monitor and control the actions of AI agents. It effectively acts as a firewall for agentic intent, ensuring that if an agent tries to execute a destructive command or access unauthorized data, the platform can block it in real-time. By supporting the Model Context Protocol (MCP), Teleport has integrated itself into the stack that developers are using to connect LLMs to external tools and data sources.
Teleport is technically a competitor to legacy VPN providers and PAM vendors, but its approach is closer to modern Zero Trust. The platform assumes that the network is compromised and relies entirely on the cryptographic identity of the requester. This eliminates the need for a corporate VPN. Instead, users and agents authenticate once through an SSO provider and receive a certificate that gives them access to a specific subset of resources for a limited time.
The business model is usage-based, charging per monthly active user or protected resource. This reflects the reality of modern infrastructure where the number of machine and agent identities often exceeds the number of human employees. By unifying human and machine access into a single audit log, Teleport provides a clear record of who—or what—performed every action in a production environment, which is a critical requirement for industries with strict regulatory hurdles.
A unified identity layer for humans, machines, and agents to access production infrastructure.
Teleport is hiring.