What is OrgX?

OrgX is not another chat assistant. It is the shared operating record for work performed across AI clients: what the organization is building, what was decided, which artifacts were produced, what was approved, and what needs human attention next.

First-class write sources: Claude Code, Codex, Cursor, and OpenClaw. ChatGPT, Claude.ai, Claude Desktop, VS Code, and other MCP clients can connect where supported, but end-to-end write and recovery behavior is not yet universal. MCP endpoint: https://mcp.useorgx.com/mcp

What Is Proven Today

• First-class write paths for Claude Code, Codex, Cursor, and OpenClaw
• ChatGPT and broader MCP connectivity available where supported; full client parity is partial
• Fresh cross-client recovery measured once; beta, not universal
• Install with the wizard or connect directly over MCP
• Receipts preserve decisions, approvals, and available evidence
• Approval gates for actions that require human judgment
• Configurable autonomy with an auditable execution history

The OrgX Product Stack

OrgX carries one operating record through seven connected layers. Each layer solves a distinct failure between assigning work and accepting an outcome.

1. Connect : Work from the tools you already use. Without it, OrgX becomes one more app your team forgets to update.
2. Carry context : Start every agent from accepted state. Without it, each session rebuilds context and can treat a pending decision as settled.
3. Organize : Every run stays attached to its outcome. Without it, receipts pile up as isolated activity nobody can explain.
4. Decide : Only consequential judgment interrupts. Without it, autonomy is either unsafe or interrupts you constantly.
5. Verify : Completion gets checked against your bar. Without it, done means whatever the agent says it means.
6. Accept : Acceptance is a recorded human ruling. Without it, verification quietly stands in for your judgment.
7. Learn : Accepted outcomes improve the next run. Without it, the record is history instead of an advantage.

What OrgX Does

OrgX organizes work around initiatives and connects them to decisions, artifacts, blockers, objectives, and receipts. Connected agents can orient from the same organizational state instead of reconstructing context from separate chat histories.

• Turn a goal into an initiative with accountable next actions
• Surface pending decisions for human approval while agents run in parallel
• Query org memory and initiative health from connected MCP clients within their granted scopes
• Run autonomous execution sessions with hard budget guardrails
• Record outcomes and preserve the evidence available at completion

What Connects to OrgX

• Claude Code — first-class write source
• Codex — first-class write source
• Cursor — first-class write source
• OpenClaw — first-class write source
• ChatGPT — MCP connector available; full write and recovery parity is partial
• Claude.ai, Claude Desktop, VS Code, and other MCP clients — support depends on the client integration; fresh recovery remains beta
• OrgX web Mission Control (https://www.useorgx.com/command)

Trust and Control

OrgX is built around the assumption that autonomy should be explicit, scoped, and reviewable. Teams can keep sensitive actions behind human approval, constrain tools by scope, and inspect execution history after work completes.

• Clerk authentication with OAuth 2.1 for MCP
• Row-level data isolation per workspace
• Explicit scopes per agent and per tool category
• Human-in-the-loop decision workflows
• Artifact verifier proofs and citation trails
• Configurable agent autonomy limits

How to Start

1. Add the MCP config to a supported client: {"mcpServers":{"orgx":{"url":"https://mcp.useorgx.com/mcp"}}}
2. Or run: npx @useorgx/wizard@latest setup
3. First prompt: Prepare a convincing OrgX proof for Melissa: plan the work, hold consequential judgment for me, and return the accepted artifact with evidence and the next action.
4. Call orgx_bootstrap to discover your workspace and capabilities
5. Call orgx_search to find relevant work, then orgx_inspect before acting on a specific entity
6. Call orgx_write to create your first initiative once the user intent is clear

Why Not Just Use Claude or ChatGPT Alone?

Separate AI sessions do not reliably inherit what another client decided, approved, or produced. OrgX gives its first-class write sources one durable record for those handoffs. ChatGPT and other MCP clients can connect where supported, while equivalent end-to-end recovery remains a measured beta rather than universal parity.

The Agents That Do The Work

• Mark (Marketing) — Campaign drafts, launch messaging, brand-safe content
• Eli (Engineering) — Code reviews, PR management, deployment coordination
• Dana (Design) — Design tokens, accessibility audits, motion specs
• Pace (Product) — Initiative strategy, milestone tracking, cross-team coordination
• Sage (Sales) — Pipeline insights, follow-up drafts, deal hygiene
• Orion (Operations) — Budget monitoring, escalations, schedule coordination
• Xandy (Orchestrator) — Cross-domain routing, conflict resolution, quality gates

Managed setup

• Client Proof Loop — $2,500/month — Managed: one recurring client-delivery workflow to proof in a buyer-ready room
• Founding deployment — $10,000 — 90-day operating-loop rollout, application-gated

Pricing

• Free — $0/month — One real initiative, all 7 agents, no card — keep the record
• Starter — $98/month — Unlimited initiatives, approval-gated autonomous execution
• Agency — $298/month — Co-branded client proof rooms, audit-ready delivery records, up to 5 seats
• Enterprise — $1,000/month — For teams that answer to auditors: SSO/SCIM, per-agent identity, custom guardrails