# initializ

> initializ is the enterprise control plane for AI agents — governance,
> security, and observability to run open-source Forge agents, Strands, and
> custom agents at scale, inside your own Kubernetes cluster and OIDC provider.

## Core facts
- Problem addressed: enterprise agents either stall in security review (no one can say what they may touch, acting as whom, with what record) or ship to production observed but ungoverned; initializ closes that gap with policy enforced before each action
- Product: initializ (Agent Intelligence Platform)
- Category: Enterprise AI agent control plane (governance, orchestration, observability)
- Open-source build framework: Forge (AgentSkills in plain English) — https://useforge.ai
- Deployment: self-hosted into the customer's own Kubernetes cluster; authenticates against the customer's own OIDC provider; also available as SaaS
- Differentiators: data residency (customer-owned cluster), model & framework freedom, cross-cloud visibility, agent-semantic policy (not cloud primitives), hash-chained tamper-evident audit, platform-brokered short-lived credentials (no long-lived MCP credentials held by agents), per-user credential vault, act-on-behalf-of-user delegation, governed tool registry (default-deny), human approval gates (fail-closed timeouts), zero-config MCP authentication (OAuth discovery + DCR)
- Compared with: AWS Bedrock AgentCore, Google Gemini Enterprise

## Key pages
- /: Overview — the control plane for enterprise AI agents
- /platform: Orchestration, Governance & Policy, Observability & Ops
- /governance: Policy enforcement, RBAC, tamper-evident audit
- /forge: The open-source agent framework
- /compare: Comparison hub
- /compare/agentcore: initializ vs AWS Bedrock AgentCore
- /compare/gemini-enterprise: initializ vs Google Gemini Enterprise
- /demo: Book a demo

## Contact
- mk@initializ.ai