Kontext raises $4 million to secure AI agents at runtime
Its platform evaluates agent identity, task context and requested tools before an action reaches an external system.
Policy at the point an agent acts
Kontext checks an agent’s identity, session, task, requested tool and target resource before allowing a consequential action. Security teams can allow or deny a request, narrow its scope, require approval or issue a scoped credential, with a record of the decision.
From discovery to enforcement
Organizations can first map which agents and tools are in use, then observe behavior before setting enforcement policies. The platform combines fixed rules for hard boundaries with context-aware risk assessment. The funding will expand engineering and support continued development of this runtime enforcement system.
Use of funds
Expand the engineering team
Continue development of the runtime enforcement platform