A man sits at his desk, surrounded by computer monitors with a range of AI tools on display.

How I Actually Use AI — In My Business and at Home

In Technology by Rainer Willeke

Where AI Actually Shows Up in My Day

There’s a lot written about what AI can theoretically do. I’d rather talk about what I actually use it for. Running a one-person IT and web business, producing my own music, and managing a household full of computers and NAS units. None of it is exotic. Most of it is just useful.

Before getting into examples, it’s worth saying where I’m actually coming from. I wrote my first code back in 1980 — over 45 years ago now — and kept going through Computer Science classes in high school, building apps at home for my parents’ business and for their friends’ businesses along the way. That led to a Computing Technology certificate and coursework in multiple languages in college. I changed career paths for a while after that, but retrained myself in the early 2000s and got back into building, this time with a focus on the web. That’s been over 20 years of web hosting and site building since, and AI assistance has really just sparked a renewed interest in me going further with something I already understood well before any of these tools existed.

Content and Writing Assistance

On the web development side, I run the AI Engine plugin across my WordPress sites. It helps with first drafts, brainstorming, and getting past a blank page when I’m writing copy for a client site or a blog post. It doesn’t replace the actual writing. Everything still gets edited, fact-checked, and rewritten in my own voice before it goes live — but it’s a genuinely useful starting point when I’d otherwise be staring at an empty editor. AI Engine can also handle image generation, not just text, once it’s connected to a suitable provider’s API. That’s useful when a project needs quick visual concepts without leaving WordPress. It also includes built-in chatbot functionality, which I can add directly to a site to handle visitor questions automatically. It’s another piece of the same plugin doing double duty rather than needing separate tools.

Web Development and Site Modernization

Every new WordPress-based website I build starts out with Themeco’s Pro theme for WordPress, which includes the powerful and AI-capable Cornerstone page builder. That built-in AI assistant, CSAI, is one I use right inside the builder when I’m working on client sites. Beyond that, I’ve used a mix of cloud-based and privately-hosted AI tools to accelerate solving a specific kind of problem that comes up constantly in this work. That problem is replacing an obsolete or unsupported WordPress plugin without breaking the site around it. Those same tools have helped me improve site UX in countless smaller ways too. They’re the kind of fixes that are individually minor but add up to a site that actually feels right to use.

A couple of recent, more ambitious examples: I built a custom plugin for a suite of locally-themed games on aboutnanoosebay.ca, a community history and nature site for my area, and a separate custom plugin for prtklmusic.net, a space-themed musical artist website. AI-assisted development sped up both builds considerably — going from concept to a working feature much faster than coding each one entirely from scratch.

AI as an SEO Assistant

AI has become a genuinely useful assistant on the SEO side of my work, too. A lot of SEO is repetitive and tedious by nature: keyword research, meta description drafts, technical audits, and the like. Carefully supervised AI assistance can accelerate that work considerably. That matters directly for clients. Less of my time goes toward the mechanical, repetitive parts of the job, and more of my valuable, billable time goes toward the judgment calls that actually move the needle on their results.

Image and Video Generation

I run ComfyUI locally for image and video generation. That includes an ongoing experiment training a model on my own likeness for a virtual-influencer style project. It’s still very much in the exploration phase. But the underlying tools — local, open-source image generation running on my own GPU rather than a subscription service — are already solid enough to be useful for things like concept art and quick visual mockups.

Organizing Photos and Files

On the practical, unglamorous side: my QNAP NAS runs built-in AI-powered tools for facial recognition, object detection, and duplicate-photo cleanup. After years of photography, I’ve got a growing library of images. Software that can actually find the duplicates and near-duplicates automatically — rather than me scrolling through folders by hand — is one of those quiet, unglamorous wins that ends up saving real time.

Why I Run Some of It Locally

Not everything needs to go through someone else’s cloud service. I run Ollama and LM Studio on one of my always-on PCs, which lets me experiment with local language models without sending anything offsite. For a small business handling client data, that distinction actually matters: a local model can’t leak what it never received. I’m also building a small local tool of my own: a digital-estate curator that uses a local model to help organize files and documents. I wanted something that keeps everything on my own hardware. On the music side, I’ll also lean on a local model as something close to a glorified dictionary or thesaurus when I’m writing lyrics. Nothing more sophisticated than that — just a fast way to find the word I’m actually looking for.

AI Narration for Video

On the content-production side, AI narration isn’t limited to the Tech Tips shorts I put out under the Raindog Solutions channel. I’m currently using it for a full slate of videos supporting EmoDyn, my physical modeling synth project. Teaser, intro, demo, and tutorial videos are all in progress. It’s a practical trade-off. It lets me turn around video content without needing a full voiceover session every time a script changes, while I still write and direct every one of those scripts myself.

AI as a General Problem-Solving Tool

Sometimes AI’s best use case has nothing to do with a specific piece of software. It’s just a genuinely useful research and diagnostic partner, and that shows up in surprisingly ordinary places: everything from gardening advice to automotive diagnostics.

The Everyday Example: My Garden

This past year was my first attempt at a deck garden, and I leaned on AI for guidance on soil, watering, and pruning as I went. The result was the sweetest, juiciest cherry tomatoes I’ve ever eaten, homegrown for the first time. Next year I’m hoping to track down some catnip seeds to grow alongside everything else, purely to keep the resident cat extra happy.

The Technical Example: A Failing Fuel Pump

On the more technical end, the clearest example I can give involves my 2002 BMW M3. It developed a recurring check engine light I suspected was either a fuel pump or injector issue. I took it to three different shops, and each one happily charged a diagnostic fee to tell me the fuel system was “fine.” To be fair to them, my car carries a long list of performance modifications. It’s entirely possible none of them had past experience with my exact scenario.

After getting nowhere — and spending real time and money in the process — I tried a different approach. I fed a freely available AI assistant (Microsoft Copilot, in this case) my car’s full specifications, including every modification. Alongside that, I gave it a detailed description of the symptom: its frequency, what seemed to trigger it, and everything else I’d observed. It narrowed the likely cause down to the fuel pump failing under specific heavy load conditions. A few follow-up questions later, I ordered a new pump — and that solved it. If you’ve read my write-up on the car’s flex fuel build, that’s the same fuel pump story; the diagnosis is just the part I hadn’t told yet.

AI That Operates Other Software Directly

There’s a newer capability worth watching too. Tools like Claude and OpenAI’s models are increasingly able to operate a web browser directly, handling common, repetitive tasks the same way a person would click through them. It’s not limited to browsers, either — even DaVinci Resolve Studio 21, my video editor of choice from Blackmagic Design, is now remote-controllable this way. I haven’t built that into my regular workflow yet. But it’s clearly the direction all of this is heading: less copying and pasting between an AI chat window and the actual software, more of the AI just doing the click-through work itself.

The Actual Takeaway

None of this is about AI replacing the work. It’s about using the right tool for the specific job in front of me — sometimes a cloud service, sometimes something running entirely on my own hardware — the same way I’d choose any other piece of software. That’s really the only practical framework that’s mattered to me. Does this tool actually save time or improve the result, and is it the right tool for where the data needs to live?