This is the start of a series on using AI as a product professional, maker, author, and journaler. I’ve been using gen-AI quite a while, but started to use it ‘for real’ with Claude Code and the possibilities it makes possible. I will continue learning in public, building stuff and sharing as I go.
I’ve been (obviously) interested in and excited about (generative) AI since the end of 2022, early 2023 – when OpenAI launched the famous ChatGPT. Especially in the last 2 years, I’ve been using AI more actively in my work, in my life, and for work. I had a few gigs (such as AI-powered company valuations, and AI-powered, personalized online learning platform), as well as a few side projects I vibe-coded in Lovable: minimalist time racker, trackwell, cost tracker, Budgette, and AI-powered book recommendation system, Pagey books. Most recently, I made a mindful journaling app micromemo. I’ve used n8n to automate stuff, and I’ve intensively used (paid) ChatGPT and Gemini.
All these have been exciting experiences, and I’ve considered myself an active – maybe even a bit of an advanced – user. However, I was mostly an end-user of different chat interfaces and web apps. Even with vibe-coding, I did make (relatively) elaborate tools but it was mostly chatting*1, but I haven’t made a ‘proper’ deep dive with the technology. That is, until I installed Claude Code.
Over the first week of February, I finally broke this barrier by playing with Claude, which is, in my honest opinion, a ‘real’ agentic AI. That is, a purpose-built agentic tool with full local workspace awareness, multi-step planning and ‘reasoning’, sub-agent architecture, deep integration… versus a general-purpose chatbot like Gemini or ChatGPT.
Claude Code was first and foremost made for developers. Its primary interface is Terminal or an Integrated Development Environment (IDE), It’s tech-heavy and not very user-friendly for the general public. Still – it lets you do, oh, so much more. Is the steep learning curve worth it? I would say it is!
Off-topic: Do we have to learn more tech-intensive tools like Claude Code as knowledge workers?
Is Claude Code something we must master? Yes and no. I’ll explain why.
No, because the tools are evolving, and we will gradually see more AI-powered features within ‘traditional’ tools, as well as more user-friendly AI-native tools. For example, now (Feb 2026) almost any web meeting tool has recording and transcription capabilities. Most of them also have AI-enabled summaries, action items etc. Some time ago, this was not the case. If you were tech savvy, you may have had your ‘own’ flow with transcription, summary, and reporting. You may have had a meeting recording agents/systems like Otter.ai. Today, it’s built-in. Otter tries to differentiate through additional capabilities, but the ‘traditional’ tools are catching up. Maybe you have your ‘own’ flow to store these transcripts and search them through natural language? I would bet Google or Microsoft will make it possible sooner or later (if not already possible).
Yes, because you will gain technical skills, be on the bleeding edge of what’s currently possible, and have fun.
Personally, I’m working on being more technical, and I work as PM on complex tech products, so it was a no-brainer: I have (and like to) to stay sharp and up to date. Think of the pros and cons and make an informed decision for yourself.
Working with Claude Code, I feel like I’m ‘starting for real‘. I will document the journey and learn in public. So stay around. Next up, I’ll write about how Claude changed my daily journaling practices and helps me with insights and more.
A Bit of a Precaution
I like to learn hands-on, oftentimes ‘test in production’*2, and break stuff. Before you start exploring AI and new technologies, understand the risks.
AI is great at scale and will execute fast – this also means making unwanted/unintended changes to a mass of your files, exposing private data, deleting stuff and so on.
Personal example: When I wanted to be more hands-on with Lovable, I integrated it with GitHub, so that the code is actually stored over there. The repository was under my company account (which I later forgot). Becoming more mindful of my expenses, I decided to slash the GitHub subscription and delete the company account. This deleted the repository, and because all the code was gone: I can’t make any change in Lovable anymore, I can’t remix the project and all my work on Pagey Books is frozen in time. Worse can happen (It could have deleted everything), but this is quite sad: neither Lovable nor Github could help bring the repo back, and the website is effectively frozen. I don’t offer paid plans etc. anymore, but they are mentioned, and there are some small errors like that plauging Pagey that I can’t fix.
Now, imagine if Claude deleted all your journal entries, or if your API keys got exposed, so you end up with $55k Google Cloud bill.
Be very careful.
Final shout out – I highly recommend Teresa Torres’ blog post How to Use Claude Code Safely: A Non-Technical Guide to Managing Risk, and her blog series on Claude in general, especially for product professionals.
Next up: The first skill I built for Claude and how it drastically created value for my journaling.
* Featured image made with GenAI & Canva
*1 I played a tiny bit with having Lovable connect with GitHub when I was building Pagey, but I haven’t used Cursor yet, and with GitHub, I had a mishap (described in the post)
*2 Potential partners/employers: Don’t worry, I’m talking only for personal projects, when ‘appropriate’ 🙂