Deploying your first agent

Before you start

You need a Sokko account on a paid plan. Your organization owns billing and the agent instances that run inside it. If you signed up through the homepage you already have a default organization, so sign in and you'll find it in the dashboard sidebar. Creating your first agent from there takes a couple of clicks.

Prefer the command line? Install the sokko CLI (Node 18+) and log in:

npm i -g sokko
sokko login

npm i -g sokko
sokko login

The login command asks for an API key you mint in the dashboard under Settings → API Keys. Either path does the same thing, so use whichever you like.

Step 1: pick a runtime

A runtime is the agent that runs inside your instance. Sokko ships four:

Runtime	What it is
openclaw	Gateway assistant reachable across chat channels (WhatsApp, Telegram, Slack, Discord), with 1,000+ community skills
hermes	Nous Research's self-improving agent: persistent memory plus a skills library that sharpens after each task
paperclip	Multi-agent orchestrator: an org chart of agents with roles, goals, budgets, and a full audit log
opensre	AI site-reliability agent that watches your systems and responds

You choose one when you create the agent. In the dashboard it's a dropdown; on the CLI it's the --runtime flag.

Step 2: create an instance

An instance is one running copy of a runtime. In the dashboard, pick a runtime, give it a name, and paste your model provider key. Sokko spins it up and shows the agent's URL, like https://nightly-refactor.sokko.ai, once it's ready.

Prefer the CLI? The same thing in one command:

sokko instances create \
  --runtime openclaw \
  --name nightly-refactor \
  --byok --provider anthropic --api-key sk-ant-...

sokko instances create \
  --runtime openclaw \
  --name nightly-refactor \
  --byok --provider anthropic --api-key sk-ant-...

That URL is private by default. Only you and the people you invite in your organization can reach the agent, never the public internet.

Step 3: watch it work

Open the instance's page in the dashboard to check status, watch streamed logs, and use a browser terminal any time. On the CLI, sokko instances ls lists the same status.

Your agent's files and memory are saved automatically and survive restarts and redeploys, so nothing is lost when it restarts.

What's next

Once you're comfortable with one instance, the same flow scales to a fleet of them. A good first upgrade is to connect your agent to Telegram so you can talk to it from your phone. And if you're curious why an always-on home matters so much in the first place, how to run an AI agent 24/7 walks through the options. The next post covers RBAC: how to let teammates manage these instances with owner, admin, and dev roles instead of sharing one login. The quickstart covers this same flow in the docs.