What is Agent 37 Cloud?

Agent 37 Cloud is a managed hosting platform built specifically for persistent AI agents like Hermes, OpenClaw, and Claude Code. Instead of running these agents on your own Mac minis or VPS setups, Agent 37 lets you provision always-on agent instances for each customer with a single API call. Each instance keeps its files, memory, and connected tools intact across sessions until you explicitly delete it. Pricing starts at $3.44 per month per customer instance, billed hourly from a prepaid balance.

Who it's for

• Founders shipping vertical agent products – Build and deploy customer-facing agent experiences without managing server infrastructure or writing integration glue code.
• Developers tired of stitching together VPS setups – Replace the complexity of renting boxes, running containers, tracking state, and streaming responses with one POST request.
• Teams needing white-label agent infrastructure – Give each customer their own persistent agent instance under your brand, with kernel isolation and spend caps on every instance.

Key features

One API call provisions an always-on instance

A single POST request creates a persistent agent instance for each customer. The response returns a routable URL at https://{id}.agent37.app that works behind your API key or a signed URL. No reverse proxy, DNS setup, or SSL certificates required.

Pay for compute, not seats

Pricing is based on the resources each instance consumes: $0.80 per vCPU, $0.70 per GB of RAM, and $0.09 per GB of disk per month. Unused prepaid time returns to your balance when you delete an instance, prorated to the hour.

Bring your own models or use theirs

You can either bring your own API keys for any model or meter model usage from the same prepaid balance. Per-instance spend caps give you full cost control across your customer base.

White-label by default

Your customers never see Agent 37. You hold the API key, and every instance runs under your brand with kernel isolation between tenants.

What stands out

"One instance is one customer, billed hourly from a prepaid balance. Less than a basic VPS, with the agent, API, and routing included."

Agent 37 eliminates the operational overhead of running persistent agents at scale. Instead of provisioning and patching stateful servers, you get a clean REST API with streaming events, scheduled runs, and built-in integrations for Gmail, Slack, Notion, and over 1,000 apps. The platform wraps each agent's CLI or SDK so you never write or host the glue code yourself.

Worth checking out if…

You're building a product that gives customers their own persistent AI agent and want to skip the server management entirely. Agent 37 is especially useful if you need white-label infrastructure, per-customer spend controls, and the ability to embed live terminals, desktops, or file browsers through signed URLs.