Àlex GuardiolaÀlex Guardiola · building in public
AI & Automation Engineer Systems that run on their own
- Personal assistant (Hermes)RunningHermes Agent · LiteLLM
- Siri voice bridgeRunningSiri Shortcuts · n8n
- Self-running homelabDownDocker · Healthchecks
- Ops agentRunningMCP · LangGraph
- Price watcherRunningPython · SQLite
I work in IT at a logistics company. At home I run a homelab with AI agents and automations I use every day, and I document how I build them.
Running Systems and real case studies
- Siri voice bridgeI ask the homelab from my watch, and it cannot touch anythingRunning
- QRFleetFrom WhatsApp reports to PDF inspections, in production
- Ops agentRead-only MCP and a diagnosis that costs centsRunning
- Power Platform in logisticsFrom copy and paste to reviewing results
- Self-running homelabBackups, alerts and updates that run on their ownDown
- HermesA Telegram assistant with my documents and my homelabRunning
Build log The latest I have written
I talk to my homelab from my Apple Watch (and it is not allowed to do anything)
A Siri Shortcut, a Raspberry Pi, n8n and an agent: how I ask my homelab questions from my wrist for 0.2 cents, why the voice path is read-only, and what an independent security review found before I open-sourced it.
An ops agent for my homelab: MCP, LangGraph and a gateway that keeps the bill honest
How I made my self-monitoring homelab answerable: a read-only MCP server, a LangGraph diagnosis agent that cannot execute anything, models routed by evaluation, and the numbers (and cost mistakes) behind it.
A homelab that runs itself: monitoring, backups and updates on a Raspberry Pi
How a Raspberry Pi and a desktop became a self-monitoring homelab: VPN-only access, a daily doctor script, 3-2-1 backups with a tested restore, and controlled updates, plus the incidents that taught me the most.