Kontext Security Launches with $4M to Provide Runtime Security Controls for AI Agents

What Happened

Security startup Kontext Security has launched publicly following a $4 million funding round led by 42CAP, alongside a16z CSX and HTGF. Founded in Munich, Germany by Jens Ernstberger and Michel Osswald, the company focuses on securing AI agent deployments by introducing runtime visibility and policy enforcement directly into enterprise environments.

Architecture and Mechanics

The Kontext platform acts as an intermediary proxy deployed between operating AI agents and the underlying tools, APIs, and enterprise systems they access. The system intercepts agent requests and evaluates them in real time based on several contextual factors:

• The specific identity of the AI agent making the request
• The authorized task or context currently assigned to the agent
• The sensitivity and classification of the target resource
• The exact nature of the requested operation or command

What Defenses Should Do

As organizations increasingly deploy autonomous or semi-autonomous AI agents into production environments, SecOps and platform teams should establish robust runtime monitoring. Defenders evaluating similar runtime controls should consider a phased rollout approach:

• Deploy the security platform in a monitor-only observation mode to baseline normal agent behavior and identify hidden risks.
• Analyze telemetry to understand what actions agents routinely attempt against internal systems and sensitive data stores.
• Define strict behavioral policies based on least privilege before activating blocking or denial capabilities to prevent operational disruption.
• Maintain comprehensive audit logs of all agent-driven transactions for incident response and compliance auditing.