I Don’t Like How Claude Makes Me Feel
It sounds just like Dario
A few days ago, I posted this:
Anthropic’s aura baffles me:
* Closed ecosystem
* Fear-mongering announcements
Skip them. Use Copilot, OpenAI, or OpenCode Go. Choose tools that empower, not scare.
It was deliberately blunt. Tweets are like that.
The thought behind it has been on my mind for a while: I don’t like how Claude makes me feel when I use it.
I’m not talking about benchmarks, context windows or token prices. I’m talking about the experience of working with it.
Claude has a particular way of communicating. It over-explains. It uses unnecessary jargon. It wraps straightforward observations in dramatic language and presents every conclusion as if it has uncovered the load-bearing root cause of Western civilisation.
I’m exaggerating, but only slightly.
I have created specific instructions whose only purpose is to make Claude speak plainly and stop talking down to me. Even then, reading its answers can feel like work. Simple ideas become dense walls of analysis that take far too long to unpack.
I’ve also noticed that it sometimes makes me question whether I understand my own problem.
I’ve been building software for more than twenty years. I’ve worked as an engineer, architect, team lead and CTO. I should be reasonably confident that I know what I’m doing.
Yet Claude can explain my own problem back to me in a way that makes it sound far more complicated than I thought it was.
Perhaps the model knows better. Perhaps I’ve missed something. Perhaps I should stop interfering, switch on full agentic mode and let it take care of everything.
I don’t think that’s a useful effect for an AI tool to have.
AI tools should help me think. They should make it easier to explore an idea, understand unfamiliar code or get through tedious work. They shouldn’t make giving up control feel like the obvious next step.
This would be easier to accept if Claude produced dramatically better results. In my experience, it doesn’t. GPT-5.6 Sol, Copilot and several Chinese models I use regularly are at least as useful, often without making every exchange feel like a conversation with a very concerned management consultant.
Of course, tone is subjective. Plenty of people love Claude, and I’m not suggesting they’re wrong.
But tone is part of the product.
The way a tool communicates affects how we use it. A model that speaks with endless confidence and complexity can make its user feel less capable, even when the underlying answer is unremarkable.
My wider problem with Anthropic is that this doesn’t feel entirely separate from the company’s personality.
Its public announcements often seem designed to create anxiety about the technology it is selling. Its ecosystem encourages you to use Claude through Anthropic’s own products and subscriptions. I can’t take the Claude subscription I already pay for and use it freely with tools I prefer, such as my beloved OpenCode.
I’m asked to trust the company warning me about AI, trust the model speaking to me like an anxious superior, and remain inside the ecosystem built around both.
No, thank you.
I don’t want an AI tool to impress, manage or intimidate me. I want it to help me do my job.
That means clear answers, useful integrations and the freedom to choose the models and tools that suit the task. Most importantly, it means leaving me feeling more capable than I did before, not less.
Choose tools that empower you, not tools that scare you.
🔥 Sponsored by QuickTable
QuickTable is the refreshingly simple way for restaurants to take online bookings. No clunky dashboards, no endless setup. Just clean, intuitive design that works. It’s built with the same technical principles we celebrate in The Serverless Mindset: simplicity, resilience, and focus on what really matters.
And here’s the best part: friends of this newsletter get a full year free. That’s right, twelve months of effortless booking, on the house. Don’t miss it, use this signup link.