Skip to Content
View all events

Building Integrated AI Agents with OpenClaw

Published by O'Reilly Media, Inc.

Intermediate content levelIntermediate

Architecture, deployment, and proactive automation

Course Outcomes:

  • Understand OpenClaw’s architecture and design philosophy
  • Deploy, customize, and extend a multichannel AI agent
  • Implement security guardrails for autonomous agents
  • Build proactive automation and multi-agent workflows

AI engineer Sajal Sharma provides a timely and comprehensive architecture-first introduction to OpenClaw, the open source framework for building self-hosted AI agents that operate across the messaging channels you already use, including WhatsApp, Telegram, Slack, Discord, and more. You’ll start by understanding the evolution from simple chatbots to persistent, integrated agents then dive deep into OpenClaw’s Gateway architecture, session model, and agent runtime, gaining enough understanding of the system’s internals to build your own version. You’ll learn how to deploy OpenClaw to a cloud VPS, giving you an always-on agent accessible from anywhere.

From there, you’ll discover how to make the agent yours: customizing its personality and behavioral boundaries through workspace bootstrap files, building custom skills from scratch, and configuring model providers. While OpenClaw ships as a personal assistant out of the box, it’s configurable enough for building any kind of agent including coding agents, analysis agents, customer support agents, and more. You’ll build proactive automation into the agent, using cron jobs, webhooks, and event-driven triggers to create workflows such as daily briefings, automated code reviews, issue triage, and multistep automations that run without any manual prompting. Finally, Sajal explains OpenClaw’s security guardrails, which are critical to safely operating an autonomous agent that has real tool access on real systems

This live event is for you because...

  • You’re a software engineer who wants to build and deploy a personal AI assistant that integrates into your existing communication channels.
  • You’re an AI/ML engineer exploring the agent deployment space and want to understand how a production self-hosted agent framework is architected.
  • You’re a product manager with a technical background who’s interested in understanding multichannel agent architectures, deployment trade-offs, and the security considerations involved.
  • You’re a technical power user who wants a privacy-respecting, self-hosted AI assistant that you fully control.

Prerequisites

  • Node.js 22+ installed on your machine
  • An API key from a supported LLM provider (Anthropic recommended)
  • A code editor (VS Code recommended) and terminal access
  • Comfort with the command line and basic terminal operations (navigating directories, running commands, editing config files)
  • Basic understanding of LLMs and AI agents (what they are, how they work at a high level)
  • Familiarity with concepts like API keys, JSON configuration, and client-server architecture
  • Node.js/TypeScript knowledge (helpful but not required)

Recommended preparation:

  • All configurations, example skills, and exercise templates demonstrated in the hands-on sections will be available in a GitHub repo shared before the course

Recommended Follow-Up:

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

The integrated AI agent and OpenClaw architecture (70 minutes)

  • Presentation: The evolution from chatbots to persistent, integrated agents; why self-hosted, multichannel agents are the next frontier; introduction to OpenClaw and what differentiates it; alternatives to OpenClaw; the Gateway control plane, channel adapter pattern, agent runtime, workspace contract (bootstrap files), session model, and companion nodes
  • Hands-on exercises: Install OpenClaw locally, start the gateway, explore WebChat, observe the agent loop in verbose mode, and inspect the workspace directory
  • Q&A
  • Break

Deployment options and going live (70 minutes)

  • Presentation: Deployment trade-off matrix across five options (local machine, dedicated home server, Docker, VPS/cloud, Nix); comparing always-on availability, local app access, cost, complexity, and security posture; remote access patterns (Tailscale Serve and Funnel, SSH tunnels); how deployment choice affects threat model
  • Hands-on exercises: Deploy OpenClaw to a cloud VPS end to end (install, onboard, start gateway, and verify remote WebChat access)
  • Q&A
  • Break

Tailoring your agent (80 minutes)

  • Presentation: The workspace as the agent’s brain; bootstrap files (SOUL.md, IDENTITY.md, USER.md, AGENTS.md, TOOLS.md) and how they shape agent behavior; the skills system architecture (bundled, managed, workspace tiers); creating custom skills from scratch; plug-ins, MCP support via MCPorter, and model configuration (including local models)
  • Hands-on exercises: Customize agent personality via SOUL.md and IDENTITY.md; create a custom workspace skill from scratch; test it live via WebChat
  • Q&A
  • Break

Security, automation, and real-world workflows (80 minutes)

  • Presentation: Security guardrails for autonomous agents (DM pairing, allowlists, sandboxing, least-privilege patterns, exec approvals, and config auditing with openclaw doctor); proactive automation with cron jobs, webhooks, and Gmail Pub/Sub; multi-agent routing and multichannel deployment patterns; real-world case study of instructor's daily-driver OpenClaw setup
  • Hands-on exercises: Configure security settings (allowlists, exec approvals, and doctor audit); set up a recurring cron job (e.g., daily briefing) with channel delivery
  • Q&A

Your Instructor

  • Sajal Sharma

    Sajal Sharma is an AI engineer and technology leader with over eight years of experience in AI/ML, specializing in natural language processing. He works at Menyala, a venture studio in Singapore, where he focuses on building AI-first products and shaping technology strategies. Previously, he led AI initiatives at various consulting and product companies, developing innovative AI solutions across industries. His on-demand course, Building AI Agents with LangGraph, is available on the O’Reilly learning platform. Sajal has delivered a guest lecture at Yale University and has been a mentor for Udacity and the University of Melbourne, where he guided students in machine learning and AI. He holds a master’s degree in information technology from the University of Melbourne.

Skills covered

  • AI Agents
  • Large Language Models (LLMs)
  • Artificial Intelligence (AI)