This session is a practical, step-by-step walkthrough to help you install OpenClaw, launch the local dashboard, connect a model, and start running your first agentic workflows. We’ll cover the main installation paths (one-command, Docker, manual/dev mode), when to choose each, and the common pitfalls that cause setup failures (ports, permissions, antivirus, localhost access). By the end, you’ll have a working local environment and a clear mental model of how OpenClaw executes tools and orchestrates actions.

What you’ll gain

• A clear decision guide: One-command vs Docker vs Manual (what to use and when)
• A working local setup: install → dashboard → model connection → first run
• Practical troubleshooting: 127.0.0.1 issues, blocked scripts, ports, firewall/AV
• Confidence in the basics: tools, permissions, and safe local execution
• A “next steps” blueprint to extend into real workflows (RAG, automations, multi-step agents)

Who it’s for

• Developers and AI builders who want to run agentic workflows locally
• Data scientists / ML engineers experimenting with tool-using agents
• Beginners who installed once and got stuck (errors, localhost refused, blocked scripts)
• Teams evaluating OpenClaw for internal prototypes (secure, local-first setup)

Prerequisites

• A laptop (Windows/macOS/Linux) with admin access
• Basic comfort with Terminal / PowerShell (copy-paste commands is enough)
• Stable internet for installation + model setup
• Optional (nice to have): Docker installed (only if you choose the Docker path)
• Optional: API key for a hosted model (if you’re not using a local model)