OpenClaw at home: from football pools analysis to NAS monitoring

An assistant that works while you're away

I've been using OpenClaw as a personal assistant for a few months now, and what surprised me most isn't what it does when I ask for something, but what it does when I don't ask for anything. The combination of cron jobs, web access, vision models, and persistent memory makes it possible to set up automations that run completely on their own in the background.

Beyond the technical and development use cases I already talked about in earlier posts, here are the personal ones I'm getting the most value out of.

Matchday analysis for the football pools

Every week, before filling out the football pools, I'd spend a good chunk of time checking standings, form, head-to-head records, injuries, and home advantage. Now OpenClaw does that analysis for me.

The setup is simple: a cron job that runs on Thursday morning and does the following:

• It goes through sports data sources (scores, stats, injury news) for every match that week.
• It gathers the relevant data: table position, form over the last five matches, head-to-head history, confirmed absences, and home advantage.
• It cross-checks all that information and generates a reasoned prediction for each match with a confidence rating.
• It sends me a summary on Telegram with the predictions, highlighting the matches where it sees a clear edge and the ones it considers unpredictable.

Does it get it right every time? Obviously not, football pools are football pools. But the analysis is far more methodical and consistent than what I used to do by eye, and it saves me an hour a week of hunting for data across different sites. I can also ask it to tweak the criteria: give more weight to home advantage, prioritize recent form, or be more conservative with draws.

Monitoring the cameras at home

I have a couple of IP cameras at home that record continuously and save snapshots to a directory on the NAS. The problem is that reviewing the footage is tedious, and the default motion alerts are pretty useless: they go off because of shadows, the neighbor's cats, or any change in lighting.

OpenClaw, with access to the file system and a vision model, gives me a much smarter alternative:

• A cron job checks new snapshots every thirty minutes.
• It runs them through the vision model with a specific prompt: "Are there people, unknown vehicles, or unusual activity in this image? Ignore animals, lighting changes, and vegetation movement."
• It only alerts me when it finds something that's actually worth paying attention to: a person at the entrance, a car it doesn't recognize parked in front of the door, or a package left at the front door.
• It includes the snapshot in the notification with a description of what it saw and why it thinks it's relevant.

The false positive rate has dropped drastically compared to the cameras' built-in motion detection. And the best part is that I can tweak the criteria just by talking to it: "from now on ignore Amazon delivery drivers too" or "also let me know if you see the garage door open".

NAS monitoring

My NAS is the nerve center of the house: backups, media, cameras, downloads. If it goes down, everything feels it. I used to have a Bash script that checked disk space and not much else. With OpenClaw, monitoring is much more complete:

• It checks the disk health using SMART data and alerts me if any indicator gets worse (reallocated sectors, temperature, power-on hours).
• It checks available space, not just free space, but the growth trend too, so it can estimate when it will fill up at the current rate.
• It verifies that backups have run correctly by checking file dates, sizes, and checksums.
• It confirms that running services (Docker, Plex, Syncthing, and the rest) are up and running without errors.
• It checks for pending updates to the firmware or installed packages.

It sends me a daily summary in the morning with the overall status and alerts me in real time if it detects something critical. It's like having a dedicated sysadmin for home infrastructure.

Other use cases that came up along the way

Once you have an autonomous agent with network access, file system access, and capable models, use cases start to appear on their own:

• It tracks prices on Amazon or PCComponentes and lets me know when they drop below my target price.
• Every morning it sends me a news summary from the RSS sources I follow, prioritized by relevance and filtered for noise.
• When an invoice arrives by email, it extracts it, checks the data, and files it in the right folder on the NAS.
• Reminders aren't just alarms, they're alerts with context. "Your car insurance expires tomorrow" includes a link to the comparison site and the price I paid last year.

The key: a single point of control

What makes all this really work is that OpenClaw brings everything together in one place. I don't have ten scripts in ten different places with ten different ways of notifying me. Everything goes through the same agent, with the same conversational interface and the same persistent memory.

If I want to change something, I just tell it on Telegram: "stop tracking the price of that RAM, I already bought it" or "add Wednesday's match to this week's analysis". No editing files, no redeploys, no cron tabs I forget about.

That said, fair warning: OpenClaw requires your own server and a certain level of comfort with the terminal. It's not an app you install on your phone and call it a day. But if you already have a homelab or a VPS, the initial setup effort pays for itself quickly with the number of small tasks it takes off your plate. If you want to go further, in deploying OpenClaw with Docker and Dokploy we break it down step by step. To understand the infrastructure behind it, in the infrastructure with Dokploy we cover the whole thing.

Another post in the OpenClaw series. You're coming from OpenClaw for testing and QA. To go back to the beginning, Deploying OpenClaw with Docker and Dokploy.