CreateOS is unclaimed —

CreateOS

The unified AI execution layer for the enterprise·createos.sh

CreateOS is a governed execution layer that sits between the AI agents an enterprise builds and the models they run on. It routes every request to the right model, enforces security policy before the model sees it, validates every output, and logs the full decision trail for audit. It is model-agnostic and can be deployed in CreateOS cloud, a customer's VPC, or fully on-prem, and includes components such as an AI Gateway, a Sandbox runtime, a multi-agent runtime, a CLI, and an MCP server.

What it's for

Invoice processing automationIT helpdesk automationDemand forecastingTerritory planningPipeline intelligenceAttrition predictionContract and document intelligenceReconciliationsEvidence bundle preparationCCTV compliance and ethical-sourcing oversightDeal prep and proposals grounded in CRM dataBoard packs, forecasts, and close summariesCustomer support with governed responsesInternal knowledge search from approved sourcesContinuous compliance policy checks

Features 18

CreateOS Agents: multi-agent runtime with specialist agents executing in parallel
CreateOS AI Gateway: single API/router across multiple model providers
CreateOS CLI for deploying apps and agents from the terminal
CreateOS Sandbox: isolated Firecracker micro-VM runtime for agent/code execution with per-second billing
Deploy from a GitHub repo, a Docker image, or an upload
Deployment to CreateOS cloud, a customer VPC, or air-gapped on-prem
Full audit trail with decision lineage, execution logs, and cost telemetry
Governed connectors to enterprise systems (CRM, ERP, data, identity) with least-privilege access
Human approval gates and autonomy controls (watch, suggest, execute) per agent
MCP server letting AI agents create, deploy, and manage CreateOS projects directly
Model-agnostic routing across OpenAI, Anthropic, Google, Mistral, Meta, open-source, and sovereign models
Output validation: hallucination checks, PII masking, content filtering
Prompt-injection checks and policy validation before requests reach a model
Region-aware compute with zero data retention
REST API for projects, deployments, environments, and domains
SAML/SSO and role-based access control, scoped per team and workflow
Skills: deploy apps from AI coding agents (Claude Code, Copilot, Gemini CLI, OpenCode, Codex, Amp, Kimi CLI)
Task-aware model routing with automatic failover across providers

At a glance

free trialself-hostableAPI

TypeWeb app

DeploymentHybrid

PlatformsWeb

Forenterprises, fintech companies, legal teams, healthcare operators, manufacturing teams, regulated enterprises

CompanyNodeOps, Inc. · 2023

Integrations

SalesforceSAPSnowflakeWorkdayNetSuiteOktaVantaDocuSignSlackGitHubStripeNotionOpenAIAnthropicGoogleMistralMeta

Complianceself-reported

ISO 27001SOC 2 Type II

Resources

Docs

Pricing

Pricing is usage-based with no tiers, seats, or minimums for AI Gateway (flat 2.5% platform fee) and Sandbox (per-second compute billing, $0 egress, 500 free credits to start). Enterprise deployments are priced as an annual platform subscription plus a deployment tier plus an optional fixed-scope forward-deployed engineering engagement, with costs agreed in writing on a quote basis; no numeric platform subscription price is published.

AI GatewayQuote

Access to dozens of models across nine providers, One API key, one bill, No hidden fees

No tiers, no seats, no minimums, no monthly commitment

SandboxQuote

Firecracker micro-VM platform, currently in alpha

Billed per second of vCPU and memory; pause a sandbox to stop paying for compute (pause to zero)

Enterprise PlatformQuote

Fixed pricing agreed in writing before work starts; quote-based, tailored per customer