Every few years tool fascination resets and a familiar panic shows up with it. “Learn this framework or fall behind” type posts plastered all over Twitter (X), YouTube, LinkedIn, all that. “Master this platform or get replaced” is another one of my favorites. I've been doing analytics work for the better part of a decade, and I've watched the named villain change five or six times while the actual job core stayed remarkably constant.

The current version of this feels much grander than before because the tool is genuinely good. AI can write the query, build the chart, draft the summary. AI can guide someone with no analytics experience towards a pretty good end product, which I’ve seen first hand. So the conclusion seems obvious: the people who use it best win. Except that's not what the last year actually showed. The companies that bet hardest on tool-as-replacement are actually hiring people back, and not for sentimental reasons.

This issue is about what survives the resets. Not because tools don't matter. They obviously do. But because tool fluency is the most replaceable thing you own, and treating it as your entire competitive edge is how you end up replaceable.

The thing that's help keep me relevant across every tool cycle isn't knowing the newest how. It's getting better at the why.

The Skill That Doesn’t Expire

In my decade I've gone through the migration from spreadsheets to "real" BI, the Tableau-versus-Power-BI conversations, the rise of dbt and the modern data stack, the notebook era, and now the AI layer that supposedly eats all of it. Each time, the messaging was identical: the new tool is the skill. Learn the tool, secure the career.

Here's what really happened each time. The people who had only learned the previous tool struggled. The people who understood what the tool was for moved over in a weekend. The migration was never really technical. It was conceptual, and the concepts didn't change. A bad chart is bad in Tableau, in Power BI, and in whatever an AI generates for you. Knowing why it's bad is the part that transferred.

That's the pattern under each change: tools encode answers to questions someone already worked out. The person who knows the question can pick up any tool that answers it. The person who only memorized one tool's answers is stranded the moment the tool changes. Tool fluency will always and inevitably depreciate (insert Thanos meme).

This current cycle is just the cleanest demonstration of this I've seen, because the rollback is measurable. After a year of AI-driven layoffs, about a third of companies that cut roles for AI have already rehired between 25% and 50% of them, and another 35.6% brought back more than half. The reason isn't that the AI broke, it's incredible already in its current state and just seems to get better. However, more than half of HR leaders said the AI needed significantly more human oversight and judgment than they expected, and only about a fifth said automation replaced roles without operational problems. Roughly a third said they lost critical skills and institutional knowledge they couldn't easily replace.

Read that as a message about skills, not headcount. What they automated was the output. What they lost was the judgment that made the output worth anything: knowing which number was wrong, which exception mattered, which "correct" answer would mislead an executive. The tool produced the chart. Nobody was left who knew why the chart could be misleading.

So the question isn't "am I keeping up with the tools." Anyone can prompt, anyone can learn the next new thing. The question is whether you're getting measurably better at the things prompting can't do for you: framing the actual problem before reaching for a tool, knowing when an answer is technically right but practically useless, carrying the context about your business that no model has been trained on. That's the work. The tools just change how we get from here to there.

According to a 2026 Careerminds survey of HR professionals who ran AI-driven layoffs, what share said AI successfully replaced only some tasks rather than entire jobs?

A. 21%
B. 38%
C. 52%
D. 66%

Answer at the bottom of this issue

Why Big Tech Is Quietly Rehiring After Mass AI Layoffs by Cloud Engineer Academy
A useful corrective that cuts the other way too: less than 5% of 2025's layoffs were directly attributed to AI, with "AI efficiency" often serving as cover for ordinary over-hiring corrections. Read more →

Klarna Reverses AI Customer Service Replacement by Tech.co
Klarna replaced ~700 support roles with an AI assistant it claimed handled 75% of chats, then started rehiring humans after quality dropped. The CEO's own framing is the tell: not a retreat from AI, but an admission that cost-first replacement produced worse service. Read more →

McDonald's Just Fired Its Drive-Thru AI and Is Turning to Humans Instead by Chris Morris @ Fortune
McDonald's deployed AI order-taking bots across 100 U.S. drive-throughs, then officially shut the test down and brought human cashiers back after viral videos showed failures like adding hundreds of dollars of nuggets to simple orders. Read more →

Summarize any article with VizBuddy

DuckDB #data-visualization #productivity
An in-process analytical database you can run from a notebook or the command line with zero setup. Worth including because it's the kind of tool that rewards understanding why columnar analytics is fast rather than just memorizing syntax. Free and open source.

Evidence #data-visualization #productivity
A framework for building BI reports as code (SQL + Markdown), version-controlled like anything else. A good example of the "judgment travels, tooling changes" thesis in practice: the report logic outlives the dashboard tool. Open source.

▶️ Data Storytelling Basics (in 3 Steps): How to Communicate Data and Numbers by Word Cortex with Anita (5 mins)
Anita challenges the old statement “the data speaks for itself” (it does not!). And she explains why and what we, as data professionals, can do to set up a great story using data.

According to a 2026 Careerminds survey of HR professionals who ran AI-driven layoffs, what share said AI successfully replaced only some tasks rather than entire jobs?

A. 21%
B. 38%
C. 52%
D. 66% ✅

66.1% of HR leaders said AI successfully replaced only some tasks, not entire jobs. Read more about it 📖

Newsletter publishing is hard work and it’s just me running the show here. If you ever feel like extending a thanks, idea, or insult you can do that here.
Or email me directly at → [email protected]

If you’re feeling generous, I also keep an Amazon Wishlist of books and tools I’m interested in.