Your Skills Have a Half-Life. Your Grit Doesn't.

I use Claude every day. Research, first drafts, data analysis, code. It's genuinely good at all of them. Work that would've taken a junior hire a week can now be done in an afternoon. For 80% of knowledge work, the distance between an "expert" and a "beginner with AI" has collapsed to almost nothing. And watching that happen made something obvious to me: everyone is still optimizing for the part that got easier.

Open LinkedIn and count the posts about the "top 10 skills to learn in 2026." Then count the courses, certifications, and bootcamps. People can spend six months "upskilling" without shipping one thing. We built an entire career culture around the skillset: resumes are skill lists, job descriptions are checklists, and the advice is always "learn this framework, get that certification, master this tool." Preparation keeps looking like progress even when nothing has moved.

The strange part is that skills are depreciating faster at the exact moment we're collecting more of them. Framework knowledge that took a year can be obsolete six months later. A marketing tactic that worked in January can be saturated by June. The language you mastered last year can have a better alternative today. AI accelerated that decay by 10x. Claude can code, GPT-5 can write strategy documents, Midjourney can design, and Sora can edit video. Somebody who has never done the work can approximate in seconds what took you years to learn.

That still leaves the 20% AI can't do: deciding what is worth building, shipping while the metrics stay flat, sending the 50th cold email on a Tuesday afternoon when you would rather do anything else, and putting your name on imperfect work so the market can tell you what is wrong. Research, drafting, crunching data, scaffolding code. AI can handle those. It can't supply the willingness.

So AI didn't make skills irrelevant. It made them insufficient. You still need to know things. Knowing them is becoming the easier part. Harvard Business School's 2026 AI trends report says organizations need "change fitness" more than technical skills. The ability to adapt matters more than the knowledge already sitting in your head. Skills have a half-life. Grit doesn't. In the AI era, mindset is the only asset you own that doesn't depreciate. Everything else starts to look more like a rental.

Adaptability is where the skill and the mindset meet. Somebody who can pick up a new tool over a weekend and ship something functional beats somebody with five years of experience in the old tool. Every time. The framework you mastered has a replacement coming. The channel you perfected is getting saturated. The useful meta-skill is getting competent without waiting for permission or a curriculum. One group believes they can figure it out and ships. The other waits to be taught and keeps upskilling. I would use AI to close that gap on purpose. Don't spend six months learning design before making the design. Use AI and ship it today. Don't spend three months learning analysis before looking at the data. Ask Claude and have something by lunch. Put the saved time into decisions, relationships, persistence, and taste, because those still require you.

Startup failure tests the same idea from another direction. CB Insights analyzed hundreds of post-mortems: 42% of startups failed because there was no market need, 29% ran out of cash, and 23% had the wrong team. "Founders didn't know enough." isn't on the list. "No market need" is a judgment problem. "Ran out of cash" becomes a persistence problem. "Wrong team" becomes a people problem. The biggest failures keep tracing back to how somebody decided, adapted, or kept going, not whether they knew JavaScript. Plenty of perfectly skilled people never build anything because starting is scary, failure is scary, and the ugly middle is uncomfortable.

I've seen the other side in every growth role and side project I've worked on. The people who win aren't always the smartest in the room. They're still in the room at midnight. They started without a perfect plan, shipped before the product felt ready, and sent the email before the pitch was perfect. That bias toward action matters because the information you need sits on the other side of doing something. You learn what the market wants by launching, what works by testing, and what you can handle by doing the hard part. One daily "hard thing" is a way to practice it: a cold email, a cold call, a public opinion, or a message to somebody far above your level.

And learning enough to start may take less time than the preparation culture suggests. Josh Kaufman's work on skill acquisition says you can learn the basics of almost any new skill with about 20 hours of deliberate practice. The "10,000-hour rule" is about world-class mastery. Competence can arrive quickly. Commitment is scarce. Twenty focused hours can get you far enough to put something ugly in front of other people. A landing page, cold-email sequence, blog post, feature, or prototype. Ship it this week, not next month.

Degrees make that less tidy. Around 62% of unicorn founders have post-graduate degrees, and only 4% of college dropouts become successful founders. That sounds like the opposite argument. But those founders didn't succeed because of their degrees. The degree correlates with success; the willingness to keep building in a market where 90% of startups die is causal. Steve Jobs, Bill Gates, Mark Zuckerberg, and Alex Mashrabov all dropped out. Mashrabov's Higgsfield reached $300M ARR in 11 months. Those people weren't more skilled than every peer who stayed in school. They kept going when the thing was broken.

That is also why I would kill one "preparation" habit before adding another. The unfinished online course instead of building, or the entrepreneurship book instead of doing entrepreneurship. For a week, count what you shipped and what you consumed. If the ratio is below 1:1, preparation is winning. Most people get stuck on one more course, one more book, one more certification, one more "I'll start when I know enough." You won't know enough. Start anyway.

Grit is what keeps the start going. Angela Duckworth's research at the University of Pennsylvania found it predicted success better than IQ, natural talent, and physical ability. At West Point, grit predicted which cadets would survive the "Beast Barracks" summer better than SAT scores, class rank, or fitness tests. Grittier children in the National Spelling Bee studied longer and advanced further than competitors with more natural talent, and Duckworth found the pattern across every domain she tested. The person who is still there on day 300 beats the genius who stopped on day 30. David Holz built Midjourney without VC funding, a marketing team, or a sales organization. Just years of product obsession until the market caught up, then $500M ARR. That's grit after it has had time to compound.

Discipline is what gives that time a schedule. Motivation arrives on Monday after a good weekend and is gone by Wednesday afternoon when nothing works. Discipline is the gym at 6 AM when it's cold. It's cold email number 500 after number 499 was ignored. It's publishing when you have nothing clever to say, or shipping a feature at 80% because done beats perfect. Most people wait to feel inspired. Builders work on schedule, and the difference after a year gets huge. Higgsfield shipped 4-7 feature releases each week. Not every release was inspired; the system required the output. That shipping discipline, more than one feature, drove the company to $300M ARR.

My prediction is that five years from now, the most successful people in tech won't be the most skilled. They will be the most relentless. They will treat obstacles like puzzles instead of walls and ship 100 things while everybody else is taking courses about shipping. Skills will keep depreciating. Grit will keep compounding. The hard thing you have been avoiding is where to start, not the next skill on your list.