I’m inundated in AI tools. Cursor and Claude Code at my day job, Codex for my websites, Cowork for managing this newsletter. I don’t think about it anymore. It’s like email, or Slack, or Google Docs. It’s just there. Part of the operating system of my work life.
And yet, every time I open LinkedIn, someone is posting a breathless breakdown of how they used ChatGPT to write a project brief. The comments are full of people tagging their colleagues. “You need to see this!” as if the person discovered fire. They used a text box on a website.
We need to talk about this, because the data says we should stop talking about this.
Everyone already uses AI
80% of employees now use AI tools at work. McKinsey’s latest puts it at 88% of organizations with AI embedded in at least one business function, up from 78% last year. Over half of professional developers use AI tools every single day. 92% of Fortune 500 companies have at least one GitHub Copilot user.
When 80% of people do something, it stops being interesting. The remaining 20% are the ones who need to explain themselves.
The tools are getting invisible (and that’s the point)
I’ve talked extensively about how tools like Cursor and Claude aren’t just for coders anymore. That thesis keeps getting stronger. In the last month alone, three frontier models shipped within 28 days of each other: Claude Opus 4.6, Gemini 3.1 Pro, and GPT-5.4. Each one pushed the benchmarks forward in reasoning, coding, and agentic tasks.
But the models are just the engine. The cars built on top of them matter more.
OpenAI shipped Codex as a full desktop app, letting you run multiple coding agents across projects at once. Anthropic launched Claude Code Desktop with parallel sessions, visual diff review, and connectors for GitHub, Slack, and Linear. They also shipped Claude Cowork, which brings all of that agentic power into spreadsheets, file management, and report prep for the non-technical crowd.
The design philosophy across all of these is the same: get out of the way. Sit behind the work. Become invisible. The best tools disappear into the task, and that’s exactly what’s happening here.
In 1997, 2% of the world’s population used the internet. Companies put “webmaster” in people’s titles. Super Bowl ads bragged about having a website. By 2003, Pew Research described internet usage as “a normalized part of daily life” and nobody said “surfing the web” without sounding like a grandparent. Six years from novelty to wallpaper.
AI is running that same track at triple speed. BCG’s CEO survey shows corporations planning to double their AI spending in 2026. CEOs are personally owning AI strategy at twice last year’s rate. When the C-suite treats something as infrastructure, the rest of the org follows. This is what normalization looks like from the top down.
The people who know, stopped talking
Here’s what I’ve noticed (anecdotally, but it’s consistent): the people who adopted AI eighteen months ago stopped mentioning it about twelve months ago. They folded it into the way they work the same way they folded in cloud storage and Slack and collaborative docs.
The people who adopted it six months ago? They’re the ones writing the LinkedIn posts. Give them another six months. They’ll stop too.
ActivTrak’s 2026 State of the Workplace backs this up. They found the productivity sweet spot sits at 35 to 50 minutes of AI tool usage per day (about 7-10% of working hours). That’s it. A wedge of your day, like checking email. You use it, you move on, you don’t write a post about it. The study also found the average workday shortened by about 9 minutes between 2023 and 2025, while productive hours went up 5%. The tool is doing its job (quietly).
Coding is the clearest example
Developers finish tasks 55% faster (with 40% fewer security vulnerabilities). Copilot, the silent giant, generates 15 million lines of code per day across its user base. One in ten pull requests on GitHub now involves Copilot-generated code.
Nobody walks into standup and says “I used autocomplete today.” AI code generation is reaching that exact same status. It’s a feature of the editor; it’s how code gets written now. Mentioning it in conversation adds nothing, the same way you wouldn’t announce that you used Google to find a Stack Overflow answer.
A notable exception
There is one context where talking about AI usage matters, and it’s methodology. If you’re building products with AI, writing about AI (hi), or making architecture decisions about where AI fits in a system, that conversation has a purpose. “We chose Claude Cowork for this project because the agentic capability and focus on local file work let us prototype three approaches in a secure manner” is a sentence worth saying. “I used AI to write this status update” isn’t.
The difference is talking about the tool as infrastructure versus talking about it as identity. Infrastructure conversations drive decisions, whereas identity conversations drive LinkedIn engagement. Know which one you’re having.
Models will keep improving. The cadence of frontier releases has compressed from years to months to weeks. Tooling will keep disappearing into the surfaces you already use. Your email client will draft responses. Your IDE will write functions. Your project management tool will summarize threads and run cross analysis.
The internet took about six years to go from novelty to wallpaper and AI is on a faster track. The social norm of announcing your AI usage has a shelf life, and it’s measured in months.
Stop saying you use AI. Just use it.