Maritime

Product Introduction

1. Definition: Maritime is a specialized cloud infrastructure platform designed for deploying, managing, and scaling stateful AI agents. It falls into the technical category of "agentic AI infrastructure" or "persistent agent hosting."
2. Core Value Proposition: Maritime exists to solve the immense operational complexity and cost of running thousands of isolated, persistent AI agents for customer-facing applications. Its primary value is providing a fully managed platform where each customer's AI agent runs in its own secure, stateful micro-Virtual Machine (micro-VM) that sleeps when idle and wakes on demand, all for a predictable, low cost starting at $1 per agent per month.

Main Features

1. Isolated Micro-VM per Agent: Each AI agent is deployed into a dedicated, lightweight virtual machine with its own kernel, filesystem, and memory. This provides hard security boundaries, preventing data leakage between agents. The technology is akin to containerization but with stronger isolation guarantees, ensuring one customer's agent cannot access another's data or secrets.
2. Automatic Sleep/Wake with Persistent State: Maritime's core innovation is its ability to automatically suspend (sleep) a micro-VM when its agent is idle, incurring minimal cost, and rapidly wake it (in ~1 second) when new traffic arrives. The agent's memory state, files, and environment are preserved exactly as they were, eliminating cold-start penalties and maintaining conversational context or task state.
3. Agent-First SDKs & Provisioning: Maritime provides first-class TypeScript and Python SDKs, a CLI, and REST APIs designed for programmatic agent lifecycle management. Developers can trigger the provisioning of a new isolated agent for a customer with a single API call from their backend, tagging it with their own external IDs and receiving webhooks for all lifecycle events (deploy, sleep, wake, fail).

Problems Solved

1. Pain Point: Companies building products where each end-user needs a dedicated, long-lived AI assistant face prohibitive infrastructure costs and engineering complexity. Managing thousands of always-on VMs or containers is expensive and operationally heavy, while serverless functions lack persistence and state.
2. Target Audience: The primary personas are Technical Founders and Engineering Teams at AI-Agent SaaS Startups, Enterprise DevOps Engineers deploying internal agent swarms, and Developers building custom, stateful AI applications. It's built for scale, from a single prototype to millions of deployed agents.
3. Use Cases: Essential scenarios include: Customer-Specific AI Assistants (e.g., a personal shopping agent for each e-commerce user), Persistent AI Workflows (e.g., an agent that monitors a data pipeline and maintains context over days), and Multi-Agent Systems where each agent requires a dedicated, secure environment with its own tools and credentials.

Unique Advantages

1. Differentiation: Unlike generic cloud compute (AWS EC2, Google Cloud) or platform-as-a-service (Railway, Fly.io), Maritime is purpose-built for the agent paradigm with built-in sleep/wake and per-agent isolation. Compared to other agent platforms (E2B, Modal), Maritime emphasizes a flat, predictable pricing model per agent and seamless scaling from one to thousands without re-architecting.
2. Key Innovation: The combination of lightweight micro-VMs for hard isolation with a state-preserving sleep/wake mechanism is the key technical innovation. This architecture allows for the density and cost profile of containers, the security of VMs, and the economic model of serverless—all while maintaining full agent state, which is non-negotiable for effective AI assistants.

Frequently Asked Questions (FAQ)

1. What is Maritime AI agent hosting? Maritime is a hosting platform that lets you deploy AI agents in isolated, stateful virtual machines that automatically sleep when idle and wake on demand, offering a simple, scalable alternative to managing your own VM infrastructure for agent-based applications.
2. How much does Maritime cost per month? Maritime uses a flat-fee pricing model. It starts at $1 per agent per month for the Scale plan, with each agent including 1 vCPU, 2GB RAM, and 5GB SSD. Plans begin with a free tier (3 agents) and scale up with included agent bundles (e.g., $20/mo for 20 agents).
3. How does Maritime's sleep/wake feature work? Maritime monitors traffic to each agent's micro-VM. After a period of inactivity, the VM is suspended to disk, drastically reducing cost. Upon the next API request, the VM is restored from memory in about one second, with all previous application state, files, and network connections intact.
4. Can I run OpenClaw or CrewAI on Maritime? Yes, Maritime is framework-agnostic. You can deploy agents built with OpenClaw, Hermes, ZeroClaw, CrewAI, LangGraph, or any custom AI agent stack. You push your code via Git, CLI, SDK, or Docker, and Maritime handles the environment.
5. Is Maritime suitable for enterprise on-premises deployment? Yes, Maritime offers an enterprise solution for on-premises deployment. This allows companies to run the Maritime platform on their own hardware, keeping all data within their private network while benefiting from the same agent isolation, scaling, and management features.