We build the platform that lets enterprises run AI agents the way they run everything else in production: governed, observable, and secure. Agents are defined in plain English with AgentSkills, run anywhere, and operate under policy — acting on behalf of your people, with a full audit trail.

Our work spans two connected projects: Forge, the open-source runtime for building and running agents, and initializ, the enterprise control plane that governs them at scale.

Forge

Forge is a secure, portable AI agent runtime. Define an agent in a single SKILL.md file, then run it locally, in a container, in Kubernetes, or fully air-gapped — the same agent, anywhere, with no inbound tunnels to expose.

• 60-second setup — forge init configures provider, keys, channels, and skills
• Secure by default — outbound-only connections, egress allowlists, encrypted secrets
• Portable — runs local, Docker, Kubernetes, or air-gapped with local models
• Observable — structured audit logs with correlation IDs for every action
• Extensible — add skills, tools, channels, and providers without touching core

# Install
curl -sSL https://raw.githubusercontent.com/initializ/forge/main/install.sh | bash
# Build and run an agent
forge init my-agent && cd my-agent && forge run

Forge gives teams a way to build and run agents. initializ gives the organization a single place to govern them.

It's the control plane for AI agents in production: agent identity and delegated access across your SaaS systems, policy and guardrails, role-based access, and end-to-end observability — so agents can do real work across the enterprise while staying auditable and compliant.

→ See the platform at initializ.ai

• 🌐 Platform — initializ.ai
• 📚 Forge docs — useforge.ai/docs
• 💬 Community — Discord
• ⚒️ Source — github.com/initializ/forge