Pretxt Launches AI Supervision Tool for Runtime Agent Control
Pretxt has emerged from stealth with $5M in seed funding and a purpose-built AI Supervision platform that monitors AI agent behaviour at runtime, intervening when agents deviate from expected business norms before harm occurs. This closes a meaningful gap for defenders managing agentic AI deployments, where static guardrails and permission scoping alone have proven insufficient to prevent misinterpretation-driven agent drift. The residual challenge lies in operational maturity — organisations must have sufficient agent observability and policy definition in place before runtime supervision can be calibrated effectively.