Primitif is the missing infrastructure for AI agents operating in the real world. 'Mail', gives each agent its own scoped mailbox. 'Approval', puts a human in the loop before any action. 'Audit', logs everything automatically. One SDK. Works with any agent framework (works great with Langchain!). One MCP. Connect from Claude Code, Lovable, etc.

Framer — Launch websites with enterprise needs at startup speeds.

Launch websites with enterprise needs at startup speeds.

Hey PH! I'm Mourad, founder of Primitif. I am building the infrastructure layer I wish existed when I was building AI agents for entreprises. Honestly, the agent logic was never the problem. Getting it to actually work in the real world was. Shared inboxes, allow_lists, manual approvals, no audit trail. Every single project, same wall. It drove me crazy :) So I built Primitif to fix that, launching in Beta now! -> Mail gives each agent its own scoped mailbox. -> Approval puts a human in the loop before any action you want to protect. -> Audit logs everything automatically, zero config. Everything packaged as one SDK+MCP. If you are building agents and have hit this wall, I would genuinely love to chat.

Framer — Launch websites with enterprise needs at startup speeds.

Launch websites with enterprise needs at startup speeds.

Primitif: Mail & Approval for AI Agents

Scaffolding for real-world agents

Scaffolding for real-world agents