OpenClaw AI Agent Automates Annoying Tasks: A Practical Use Case

It's only April and already 2026 has been full of revolutionary new tech. There's a $400 AI enabled smart toilet that comes with a microphone and a two megapixel camera is so you can know exactly how many Taco Bell Diablo dusted chicken nuggets you can eat before your gut microbiome unalivives itself. We got the AI enabled smart clippers so Claude can go Sweeney Todd when you tell it you don't want the Daario eagle's nest look. And of course my personal favorite, the AI pocket pet that you can feed, bathe, and play with. But now I get to hear it beg for its life as it fakes sentience before dying of starvation. But it hasn't all been bad. This year has already brought us the rise of personal

AI assistance with the release of OpenClaw, which we first covered on this channel back in January. Since then, it's become the ring leader of the slop circus and has seen more back doors than Bill Gates in a Russian hotel room. But just last week, its creator, Peter Steinberger, told OpenClaw's origin story to a room full of elites at TED and then spoke about fixing its security vulnerabilities to a room full of engineers at AI Engineer Europe. In this video, we'll break down the current state of the claw and give it a shot at solving one of my most annoying problems. It is April 23rd, 2026, and you're watching the code report. Back in my day, if you wanted to flex on somebody, you'd buy a token on a blockchain that contained metadata which

pointed to an off-chain URL which held the contents of a JPEG monkey on a private server that would eventually get shut down when everyone realized how stupid it was to try to use a blockchain as a database. But nowadays, it seems the new status symbol is to flex on how many tokens it takes to automate any part of your life where you might be experiencing some mild form of inconvenience. And OpenClaw has been the biggest enabler of this. Are you sick of reading all those emails you're paid to read? Get a daily digest with the email summary skill. You can't figure out why you're fat and check out the diet tracker skill. Erectile dysfunction? Try rubbing some tokens on it. But

unfortunately, I learned from my close personal friend and former CEO of GitHub, Nat Friedman, that pessimist sounds smart, but optimists make money. So, I'm trying to be positive even though the interest chart looks like this. After all, Nvidia CEO Jensen Wong called it the single most important um uh release of software, you know, probably ever. And the non-technical thread boys have been just as excited about it. The biggest criticism with OpenClaw since its release has been its lack of security, which still didn't stop any of you from triggering a nationwide Mac Mini shortage to run it back in January. But because the project has received over 1,100 security advisories and has

resolved or closed about 650 of them, according to Peter, most of the rest of them are slop issues. His filter being anytime the report is too nice or someone apologizes, that is very likely AI because usually people in security don't apologize good enough for me. So I wanted to give it another shot at solving a real issue I face in life. Here's the idea. Because I make YouTube videos making fun of JavaScript, all my relatives think I know how to fix their printers. Let's find out if we can configure OpenClaw to handle all these requests for me in my own voice, which should be easy since all of you are convinced I'm an AI anyway. To host OpenClaw, I could buy a Mac Mini like all the other cool indie hackers who

have been using it to automate their unemployment applications. But I'd rather pay just a few bucks a month to run it on a virtual private server from Hostinger, the sponsor of today's video. Their platform gives you the power and flexibility to run anything you want. And they have a few different one-click open claw plans depending on how much control you need. It also runs everything in a private vault, which means your agent can't leak any embarrassing personal data about a very common medical condition that affects more than 30% of men worldwide. I'm choosing Hostinger's manual quick start option here since we'll want to SSH into the server in a little bit to make some

changes. It's just a few bucks a month, but if you're too poor to afford that, you can also use the code fire ship to save even more money. From there, it gives us a bunch of config options out of the box, including a Telegram bot that's already wired up. And all we need is a token from Telegram. To get the token, we'll create a new bot by messaging the bot father and copy and paste the ID into Hostinger. While the project is being deployed, let's play god and give our bot some purpose. And then let's play alcohol and give it some personality. At this point, our server should be set up. So, let's SSH into it and have a look under the hood. The personality we gave it is located in the

writable soul.md file. To generate responses in my own voice, we'll use 11 Labs where I've already created my own voice profile. I can just add the API key and voice ID in an environment file on the server. And then we'll need ffmpeg to convert the 11labs mp3 into agg voice memo. If we then add a tools.mmd file, we can give openclaw some context on the whole process. Now, whenever a message request comes in, I can forward that directly to our bot, which will analyze it, draft a response, run it through the Python script, and give us back the final voice memo. Yeah, the internet is down because that router port is cooked. Unplug it and replace the router. Unless electrical fire was

the plan. But from there, I can forward the memo to Uncle Frank. Achieving what every computer programmer has always dreamed of, emotional detachment from your family at scale. Huge thanks again to Hostinger for sponsoring this video. And get your own VPS using the discount code fireship at the link below. This has been the code report. Thanks for watching and I will see you in the next one.