The Agent Board Guide
Building your personal AI operations team.
A lived-experience guide to standing up AI agents — written by people who built it, for people who want to.
What this guide is
A non-technical operator used Claude as a copilot to build a personalised AI agent system from scratch — hardened cloud server, customised identity files, multi-channel communication — in a focused afternoon of work. This guide is the result. Everything in it was earned from real work, including the incidents that went wrong and how they were resolved.
This isn't theory. It's a build log turned into a reusable guide.
How to use this guide
This guide is built around two main paths depending on your starting point.
Path A: “I just want to see if this works” (30 minutes)
- Tab 1 (Quick Start) — get a working agent on Telegram in 30 minutes. Prove the concept.
- If it works and you want more, continue to Path B.
Path B: “I'm building this for real” (focused afternoon)
- Tab 3 (One, Two, or More?) — decide whether you want one agent or more before you start building.
- Tab 4 (Using Claude as Your Copilot) — read this before Tab 2 if you've never used Claude Code.
- Tab 2 (The Full Build) — the main event. 7 phases. 3-4 hours of focused work.
- Tab 5 (Troubleshooting) — keep open in another tab while you build.
- Tab 6 (Models & Cost Optimisation) — come back here after you've built something and want to tune it.
- Tab 7 (Framework Reference) — your “look it up when you need it” reference.