Beginner → Intermediate
Building an AI Agent
TRAINING · FREE GUIDES
Building an AI Agent
From decision to production
Two complete step-by-step workshops: a business agent with no code, and an autonomous agent on your own server — with the security nobody explains to you.
Beginner → Intermediate · 80 min read · 42 pages
What you will learn
- The real difference between automation, copilot and agent — and why confusing them is expensive
- The anatomy of an agent (trigger · brain · tools · memory) and its reasoning loop
- WORKSHOP 1 (no-code, 9 steps): an agent that qualifies your inbound requests, scores them, enriches them and drafts your replies — with a human in the loop
- The system prompt, where 80% of the quality is decided: the full text, ready to copy
- What autonomous agents are (Hermes by Nous Research, OpenClaw) and what they change
- WORKSHOP 2 — Hermes Agent (12 steps): a complete, secured deployment — isolated machine, hardening, minimal mode rather than “allow everything”, channel locking through pairing, permissions by tool family, scheduler, memory vs skills, cost optimisation
- The real security incidents, the warning from the French authorities, and the defence against prompt injection
- The real budget, the real timeline, the 5 pitfalls, and how to know when to stop
Contents
- I. Understanding
- II. Deciding
- III. Choosing your platform
- IV. Workshop 1: the business agent
- V. The system prompt ★
- VI. Autonomous agents
- VII. Workshop 2: the autonomous agent
- VIII. Security and governance
- IX. Running it over time
- X. Going further
- Appendices: production readiness checklist · monitoring agent instructions · 30-day plan · FAQ
Bonus
2 complete system prompts (qualification + monitoring) + security checklist + 30-day plan