Most AI products still live behind a chat box on a website. Hosted Agents is different. You deploy an agent once, connect it to the apps your team already uses, and it works there: answering in Telegram, triaging in Slack, running tasks in Discord. The runtime, scaling, and security are managed by EmpirioLabs, so you focus on what the agent should do rather than on keeping a server alive.
Hosted Agents is now generally available to every account. Each agent runs one of two agent runtimes, OpenClaw or Hermes, on top of any EmpirioLabs model.
OpenClaw and Hermes
OpenClaw and Hermes are both production agent runtimes with strong tool use, memory, and channel support. You pick one when you deploy. Either way the agent is fully managed: EmpirioLabs runs it in an isolated, resource-reserved sandbox and keeps it online.
What an agent can do
A hosted agent is not just a chat wrapper. It runs code in a secure sandbox, generates media across image, video, speech, and music through your EmpirioLabs models, browses the web for current information, and calls your own tools through remote connectors. Connect Gmail, Drive, Notion, GitHub, Slack, and hundreds of other apps in a click, and the agent can act on them.
For messaging, an agent can live in Telegram, Discord, Slack, and Matrix, with Microsoft Teams, Mattermost, and WhatsApp also available on OpenClaw. Connect a channel and the agent appears under your own bot identity there.
Pick your model and provider
By default an agent uses EmpirioLabs models, and you can choose any model in the catalog. You can also point an agent at your connected Poe account or at a custom OpenAI-compatible or Anthropic-compatible endpoint. The provider and model are switchable at any time from the dashboard or the API, with no redeploy.
Deploy from the dashboard or the API
Deploying takes one screen in the dashboard: pick a runtime, a model, and a plan. The same thing runs from the API:
curl https://api.empiriolabs.ai/v1/hosted-agents/instances \
-H "Authorization: Bearer $EMPIRIOLABS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"name": "My Assistant",
"agent_type": "openclaw",
"plan_slug": "hosted-agent-basic"
}'Once the agent is running you can message it directly, change its model, manage its skills and connectors, set who is allowed to message it, and connect channels, all from the API. Send it a message like this:
curl https://api.empiriolabs.ai/v1/hosted-agents/instances/AGENT_ID/message \
-H "Authorization: Bearer $EMPIRIOLABS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"messages": [{ "role": "user", "content": "Summarize my unread email" }]
}'Pricing
Each agent is its own monthly subscription billed to your credits. Basic is 15 USD per month and Pro is 30 USD per month. The plan covers the managed runtime; the models your agent uses are billed separately at standard model rates, and everything shows up in /v1/account/usage.
White-labeling
If you want your agents to run fully under your own brand, or you need a higher agent limit or dedicated capacity, email support@empiriolabs.ai and we will set it up with you.
Get started
Open Hosted Agents in the dashboard to deploy your first agent, or read the Hosted Agents documentation for the full API.
