Maritime

Maritime is a managed cloud platform built specifically for deploying and scaling AI agents, offered by Maritime AI, Inc. Each agent runs in its own isolated micro-VM with a dedicated vCPU, RAM, SSD, and encrypted secrets — not a shared runtime. The platform targets developers, startups, and enterprises who want to ship customer-facing AI agents without managing infrastructure.

What It Is

Maritime is an agent hosting platform that provisions isolated micro-VMs on demand, one per agent or one per customer. It handles the full lifecycle — deploy, sleep, wake, scale — so teams can focus on agent logic rather than infrastructure. The platform supports popular frameworks including OpenClaw, Hermes, ZeroClaw, CrewAI, LangGraph, Claude Code, and DeepSeek Harness, as well as fully custom containers.

How the Deployment Model Works

Maritime's core design is one micro-VM per agent, not threads in a shared process. Each VM has its own kernel, encrypted secrets mounted only into that VM, and persistent state (memory, files) that survives sleep cycles. Agents wake from sleep in approximately one second with no cold-start penalty. Deployment paths include:

• CLI: maritime deploy from a terminal or from another AI agent
• GitHub: git push maritime main
• SDK: TypeScript (npm) and Python (PyPI) with a REST API underneath
• Docker: push a container image directly

The CLI is fully non-interactive, meaning agents can programmatically provision other agents via the Maritime API.

Isolation and Security Architecture

Maritime emphasizes hard isolation boundaries as a selling point for multi-tenant agent products. Each micro-VM runs its own kernel rather than sharing a process with other customers' agents. Credentials are encrypted at rest and mounted only into the specific VM they belong to. The platform also provides lifecycle webhooks so a backend knows the moment any agent deploys, sleeps, wakes, or fails. Agents are tagged with external IDs so the customer's own database remains the source of truth.

Framework and Ecosystem Support

Maritime ships with several first-party frameworks alongside support for bring-your-own stacks:

• OpenClaw — Maritime's primary agent framework, with a browser automation variant
• Hermes — a supported harness available via one command
• ZeroClaw — another first-party framework option
• Claude Code, DeepSeek Harness — documented framework templates
• CrewAI, LangGraph, any container — custom framework support via Dockerfile

The platform also includes an LLM proxy option and a startup program offering hosting and LLM credit grants.

Target Audience and Use Cases

Maritime segments its audience into three tiers on the homepage: individual developers shipping a first agent, startups selling agents to customers (one isolated VM per customer signup), and enterprises running Maritime on-premises on their own hardware. The on-prem enterprise path keeps data inside the customer's network with auto-scaling and isolation managed internally. The homepage displays logos of companies whose developers reportedly build on Maritime, including Google, Samsung, TikTok, MIT, and Harvard, though these are presented as developer affiliations rather than enterprise contracts.

Current Status

Maritime is actively available with a self-serve signup requiring no credit card for the free tier. The platform positions itself against E2B, LangGraph Platform, Modal, AWS EC2, and other compute providers on cost and agent-first design. Documentation covers quickstart, CLI reference, REST API, framework guides, and a machine-readable contract (maritime guide --json) for AI agents scripting the CLI. The company is headquartered under Maritime AI, Inc. and was active as of 2026 per the site footer.