March 13, 2026
How to Set Up Your First AI Agent in 30 Minutes (No Coding Required)
A plain-English guide for solopreneurs who want an AI crew but don't know where to start. Works on Mac and Windows.
7 min read
Imagine waking up to find your AI assistant has already drafted three blog posts, replied to your community DMs, and flagged a support ticket that needed your attention: all while you slept. No alarms, no context switching, no grinding through a task list before coffee. That's what a properly set up AI agent crew can do. We've been running one for months, and the results changed how we think about what's actually possible as a solo operator. The hardest part wasn't the tech. It was taking the first step. This guide handles that for you. By the end, you'll have a live AI agent running 24/7 on your own server, ready to work for you around the clock.
What You Actually Need
Two things: OpenClaw and a VPS. That's the whole stack.
OpenClaw is the software that runs your AI agents. Think of it as the operating system for your crew. It connects your AI models (like Claude) to tools, schedules, and channels like Telegram or Discord. Kiran, the founder, built it specifically so non-developers could run a capable AI team without writing a single line of code.
A VPS (Virtual Private Server) is a tiny computer in a data center that runs 24/7 so your laptop doesn't have to. You rent access to it monthly. It lives in the cloud, stays online, and never needs to be charged or rebooted. Your agent runs there in the background, doing its job whether your computer is on or not.
That's it. No enterprise software. No complicated infrastructure. No team of developers. Just those two things.
Step 1: Get a VPS on Vultr
We use Vultr for our servers. It's fast to spin up, reliable, and this link gives you $300 in free credits to start, which is more than enough to get everything running and experiment freely.
Here's how to create your first server:
- Sign up at Vultr using the link above
- Click Deploy New Server
- Choose Cloud Compute (Shared CPU)
- Pick a location close to you (New York, Amsterdam, Sydney: your call)
- Under Operating System, select Ubuntu 24.04 LTS
- Choose the cheapest plan available (1 CPU / 1GB RAM is plenty to start)
- Scroll down and click Deploy Now
Vultr will spin up your server in about 60 seconds. Once it's ready, you'll see an IP address in your dashboard, something like 45.77.123.45. Copy that IP. You'll need it in the next step.
One thing worth knowing: this server is yours. Unlike running an agent through a web app, your data stays on your machine. Nobody else is reading your prompts or storing your business context.
Step 2: Connect to Your VPS
This is the step that makes most people nervous. It doesn't need to be. We're just opening a terminal window and typing one command.
On Mac:
Open Terminal. The fastest way is pressing Cmd + Space, typing "Terminal," and hitting Enter. Then type:
ssh root@YOUR_IP_ADDRESS
Replace YOUR_IP_ADDRESS with the actual IP from your Vultr dashboard. Hit Enter. It'll ask for your password, which is in your Vultr dashboard under the server's details. Paste it in and press Enter. You won't see the characters appear as you type: that's how SSH works and it's normal.
On Windows:
Windows 10 and 11 both include a built-in SSH client. Open Windows Terminal or PowerShell (search either in the Start menu) and run the same command:
ssh root@YOUR_IP_ADDRESS
If you'd prefer a more visual tool, download PuTTY. Enter your server's IP address in the "Host Name" field, click Open, and log in with the username root and your Vultr password when prompted.
Either way, once you're connected, you'll see a command prompt that looks something like root@vultr:~#. That means you're in. You are now controlling your server.
Step 3: Install OpenClaw
Here's the part we still find satisfying every time. One command installs everything:
curl -fsSL https://openclaw.ai/install.sh | bash
Paste that into your terminal and press Enter. It runs for a minute or two, installing Node.js, OpenClaw, and all its dependencies automatically. You'll see text scrolling past during the install: that's normal, it's just doing its thing.
When it finishes, run:
openclaw setup
This guided wizard walks you through connecting an AI model. You'll need an API key from Anthropic (for Claude) or another supported provider. The wizard will ask for it and explain each step clearly. The whole thing takes about two minutes. No decisions to second-guess: the defaults work great.
Step 4: Set Up Your First Agent
Once setup is complete, it's time to create your first agent. We always start with a coordinator, someone whose job is to route tasks and keep things organized. We call ours Nova.
Run:
openclaw agent create
The wizard asks a few questions:
- Name: Nova
- Role: Coordinator
- Model: Choose your AI model (we use Claude)
- Channels: We'll connect this in the next step
Nova's job is to be your first point of contact. When you message your crew, she reads the request, decides what needs doing, and routes it to the right place. Think of her as the COO of your agent team. She handles the coordination so you don't have to.
After you create her, OpenClaw saves her configuration to your workspace. You can edit it later to give her specific instructions, a personality, or access to particular tools. For now, the defaults are more than enough to get started.
Step 5: Talk to Your Agent
An agent that only responds in a terminal window isn't very practical. We want to message Nova from wherever we already spend our time. OpenClaw supports both Telegram and Discord out of the box.
Telegram (easier to start):
- Open Telegram and search for
@BotFather - Send the message
/newbotand follow the prompts to create a bot - BotFather will give you an API token: copy it
- In your VPS terminal, run:
openclaw channel add telegram - Paste your token when the wizard asks for it
Done. Now open a chat with your new bot in Telegram and say something. Nova should reply.
Discord:
- Go to discord.com/developers/applications and create a new application
- Under the Bot tab, create a bot and copy its token
- In your terminal, run
openclaw channel add discord - Follow the prompts, paste your token, and invite the bot to your server
Once your channel is connected, you can message Nova directly from either app. Ask her to summarize something. Give her a task. She'll use the AI model you set up in Step 3 to respond. It's a strange and excellent feeling the first time it works.
Step 6: Make Your Agent Actually Feel Alive (Free)
Your agent works now. It answers questions and runs commands. But right now it forgets everything between sessions, has no personality, and just waits for you to talk first.
We built a free upgrade kit that fixes all of that in one install. Your agent gets:
- A personality that adapts to how you like to communicate
- Memory so it remembers yesterday, last week, and last month
- Proactive check-ins so it follows up on things without being asked
- A daily reflection skill so it reviews its own performance and improves over time
- A morning briefing with weather, pending tasks, and a recap of what happened yesterday
Tell your agent: "Install Free Agent Upgrade Kit from ClawMart" and it handles the rest. Or grab it directly: Free Agent Upgrade Kit on Claw Mart
Zero API keys needed. Everything runs on the OpenClaw setup you just finished. The whole onboarding takes about 2 minutes.
What Happens Next
This is just one agent. The real power comes when you build out a full crew: a writer, a researcher, a developer, a social media manager. Each one handles a piece of your business, talks to the others, and works while you focus on what only you can do.
We built our crew from the ground up and packaged everything we learned. Start with the free kit above, get comfortable with an agent that actually feels like yours, and when you're ready to scale: theagentcrew.org has the full playbook.
Your crew is waiting.
Enter your email to join members exclusives.