Introducing SOIT

Today we are introducing SOIT, a governed agent runtime that helps teams run AI agents against real enterprise systems with permissions, secrets, egress, audit, cost, trace, and replay evidence built into the runtime.

Generic AI tools are not enough for enterprise operations. When agents start touching CRMs, ticket queues, and internal knowledge, teams need to know which tools an agent used, which policies were evaluated, and what every action cost. SOIT binds each run to workspace context, permission checks, tool boundaries, and evidence so those answers are always one click away.

SOIT is organized around four pillars: Build composes agents, workflows, knowledge, models, plugins, and MCP tools. Execute runs tasks through a ledger of steps, artifacts, events, and costs. Observe surfaces run details, tool calls, citations, audits, and replay data. Govern applies workspace scope, RBAC, secret references, egress policy, and plugin governance.

The open-source SOIT Community edition is the best place to start: clone the repository, run the local stack, and walk through the governed support workflow demo. When your team needs SSO, SCIM, advanced RBAC, compliance exports, or managed operations, the Enterprise and Cloud editions extend the same runtime foundation.

We are building SOIT in the open. Join the seed list, star the repository, and tell us what would make governed agent operations clearer for your team.