OpenClaw Direct today launched managed hosting for Hermes by Nous Research at openclaw.direct/hermes, expanding its platform beyond a single agent harness and giving users a choice between managed Hermes and OpenClaw runtimes.
The launch follows repeated requests from OpenClaw Direct users to add Hermes support. It also addresses a wider problem facing teams adopting AI agents: choosing an agent harness should not require rebuilding the deployment, storage, access, and monitoring layers around it.
People have been asking us to support Hermes because different agents are good at different kinds of work. We do not think users should have to accept infrastructure lock-in just to get managed hosting. They should be able to choose the harness that fits the job while we handle the server work around it.
From agent experiments to dependable deployments
Demand for AI agents is rising faster than many organizations can put them into reliable production. McKinsey’s 2025 global AI survey found that 88% of respondents said their organizations regularly used AI in at least one business function, but only about one-third had begun scaling AI across the enterprise. For agentic AI, 23% reported scaling a system in at least one function while 39% were still experimenting.
Three other signals shaped OpenClaw Direct’s decision to support more than one harness.
Verification remains a burden
Stack Overflow’s 2025 Developer Survey found that 84% of respondents used or planned to use AI tools in their development process. At the same time, 46% distrusted the accuracy of AI output, and 45% said debugging AI-generated code was time-consuming. Managed infrastructure cannot validate an agent’s work, but it can remove server maintenance from the same workload.
Governance has not caught up with adoption
A 2025 Gartner survey of 360 IT application leaders found that 74% viewed AI agents as a new attack vector, while only 13% strongly agreed that their organizations had the right governance structures in place. A consistent hosting and access layer gives operators a clearer foundation for controlling how agents are deployed and reached.
Interoperability is becoming an ecosystem priority
Google introduced its Agent2Agent protocol in April 2025 with support and contributions from more than 50 technology partners. That industry push reflects a market in which multiple agents, tools, and frameworks are expected to coexist. OpenClaw Direct’s multi-harness approach applies the same principle at the hosting layer: one managed platform should not dictate a single runtime for every use case.
What managed Hermes hosting includes
Each Hermes instance runs in a dedicated managed deployment. OpenClaw Direct provisions and operates the hosting layer while preserving Hermes configuration, memory, and installed skills on persistent storage across normal runtime restarts.
- A dedicated managed Hermes runtime
- Persistent configuration, memory, and skills
- An authenticated web dashboard
- Telegram connectivity
- Managed routing to supported AI models
- Supported ChatGPT and OpenAI Codex device pairing
- Setup, billing, and instance controls in the existing OpenClaw Direct portal
Users select Hermes during the standard setup flow and use the same managed-hosting plans available on OpenClaw Direct. OpenClaw remains available for users who prefer that runtime or require its integrations, including Slack, which is not currently supported for hosted Hermes instances.
One managed platform should not dictate one runtime
The launch establishes a multi-runtime foundation. OpenClaw remains available, while Hermes users can select a different harness for workloads that benefit from its approach to agent operation, learning, and extensibility. Users can choose the runtime first and keep the operational experience in one platform.
Managed Hermes hosting is available now at openclaw.direct/hermes.
Sources: McKinsey, “The State of AI: Global Survey 2025”; Stack Overflow, “2025 Developer Survey — AI”; Gartner survey on autonomous AI agents; Google Developers Blog, “Announcing the Agent2Agent Protocol”.