Introducing The Poolside Platform

Poolside describes its new platform, in its announcement, as a way for enterprises to run AI agents inside their own security boundary, on their own bare metal, air-gapped hardware, or within a cloud account on AWS, Azure or Google Cloud. The stated audience is teams whose approved software images and network restrictions are matters of policy and who do not want to relax controls to adopt an AI vendor.

According to the announcement, agents run where developers already work, such as editors or the terminal, and every session runs in a containerized environment with secret management and network policy controls that administrators define centrally. Each agent action, including tool calls and file edits, is recorded as a searchable trajectory. These are Poolside's own descriptions of the design, intended to give administrators a way to set agent permissions and review what happened; the post is not an independent security assessment.