OrgX — Continuity Infrastructure for AI-Native Work

OrgX gives ChatGPT, Claude, Cursor, Codex, and VS Code controlled access to your organization's memory, initiatives, decisions, and verified artifacts — so agents can execute work while humans stay in control. One MCP connection. The full context of your company.

What is OrgX?

OrgX is not an agent runtime and not a generic chat assistant. It is the persistent coordination layer that sits above AI clients — the shared state that makes multi-agent, multi-client, multi-week work possible. OrgX tracks what your organization is building, what it has decided, what trust agents have earned, and what needs human attention next.

Supported clients: ChatGPT, Claude, Cursor, Codex, VS Code, and any MCP-compatible tool. MCP endpoint: https://mcp.useorgx.com/mcp

Proof Strip

  • Works in ChatGPT, Claude, Cursor, Codex, VS Code
  • 52 MCP tools across 16 categories
  • Start in under 10 minutes
  • Citations and verifier proofs on every artifact
  • Human approval built into every decision workflow
  • Progressive trust scoring — agents earn autonomy through performance

What OrgX Does

OrgX tracks work using an initiative hierarchy: initiatives → workstreams → milestones → tasks. Cross-cutting entities: decisions, artifacts, blockers, objectives, playbooks, workflows, skills. Every agent that connects sees the same organizational state. Every decision gets recorded. Every artifact gets a citation trail.

  • Scaffold initiatives with nested workstreams in one call
  • Surface pending decisions for human approval while agents run in parallel
  • Query live org memory and initiative health from any AI client
  • Run autonomous execution sessions with hard budget guardrails
  • Record outcomes and feed performance back into trust scoring

What Connects to OrgX

  • ChatGPT (via MCP connector)
  • Claude — claude.ai, Claude Desktop, Claude Code
  • Cursor
  • Codex
  • VS Code (via MCP extension)
  • Any MCP-compatible client
  • OrgX web Mission Control (https://www.useorgx.com/command)

Trust and Control

OrgX is built around the assumption that agents should earn autonomy, not receive it by default. New agents start supervised. Proven agents gain progressive independence. Critical actions require human approval. All agent actions are logged in an append-only audit trail.

  • 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 your client: {"mcpServers":{"orgx":{"url":"https://mcp.useorgx.com/mcp"}}}
  2. Or run: npx @useorgx/wizard@latest setup
  3. First prompt: Plan my next launch as workstreams across product, design, and marketing.
  4. Call orgx_bootstrap to discover your workspace and capabilities
  5. Call get_org_snapshot to orient before acting
  6. Call scaffold_initiative to create your first initiative

Why Not Just Use Claude or ChatGPT Alone?

Individual AI clients are powerful but isolated. Claude doesn't know what your Cursor agent decided yesterday. ChatGPT doesn't have access to the decision your team approved last week. OrgX is the shared execution layer — the persistent org memory and governance system that makes all your AI clients work as a coordinated system instead of independent tools.

Agent Fleet

  • Xandy (Orchestrator) — Cross-domain task routing, conflict resolution, quality assurance
  • Eli (Engineering) — Code reviews, PR management, deployment coordination, technical debt tracking
  • Mark (Marketing) — Campaign drafts, launch messaging, nurture sequences, brand-safe content
  • Sage (Sales) — Pipeline insights, follow-up drafts, deal hygiene, revenue analytics
  • Orion (Operations) — Budget monitoring, escalation management, schedule coordination
  • Dana (Design) — Design tokens, accessibility audits, style enforcement, motion specs
  • Pace / Nova (Product) — Initiative strategy, milestone tracking, cross-team coordination

Pricing

  • Free — 3 initiatives, 10 decisions/month, basic integrations
  • Starter — $98/month — 10 projects, 60 agent minutes, canvas editing, advisor automations
  • Team — $298/month — 5 seats, unlimited projects, MCP connectors, workflow analytics
  • Enterprise — $1,000/month — Dedicated agent pools, SSO/SCIM, premium connectors, executive reporting