Just-in-Time Learning: You Don’t Have to Learn Everything
Everyone’s telling you to panic-learn AI right now. After 22 years in tech, here’s why it’s okay to be late.
The loudest career advice in your feed right now is also the most wrong.
Learn everything. Learn it now. Or become a fossil.
I’ve been in this industry for 22 years, through more “this changes everything” moments than I can count, and I can tell you plainly: that’s not how staying employed works. It’s not even how staying good works.
Recently I wrote about navigating this brutal job market, and part of my advice was to keep your skills sharp and use AI sparingly. I could feel the next question forming before I’d even finished: How sharp? Sharp enough for what? And if everyone’s screaming that I need to learn AI right now or get left behind — am I already too late?
So let me take the pressure off.
A quick story about being behind
When I moved to the Bay Area in 2004, the first thing I did was nothing.
I’d just finished engineering school: four years of working my butt off, overloading on graduate classes my senior year while everyone around me coasted on senioritis. I was tired. So for the first six to nine months, I worked 9-to-5, went home, read, did a lot of yoga, and let my brain recover.
Then I started to notice things. A lot of my assignments felt rote, work I’d been handed as an intern years earlier. And the whole market was moving toward web development while I was sitting there writing C code. I got scared of the thing every engineer is quietly scared of: getting pigeonholed. Becoming the person who’s really good at one narrowing thing right up until that thing isn’t needed anymore.
Here’s what I didn’t do. I didn’t sign up for every extra project to make myself indispensable — I’d watched that backfire on people, who either burned out or got labeled as the go-to for one niche and never got to reinvent themselves. And I didn’t try to learn everything at once.
What I did was small and boring. I asked my manager if I could take some distance-learning courses Stanford was streaming. There was a room I could duck into at lunch to watch the lectures, as long as it didn’t interfere with my day job. It never did. That was the whole strategy: keep the blade sharp enough, on my own terms, before anyone could box me in.
Learn enough to stay ahead — not everything
That experience taught me the rule I still use:
You don’t need to learn everything. You need to learn enough to stay ahead, and then prove you can learn and adapt fast.
Those are two different skills, and the second one matters more. Nobody can keep up with every framework, every tool, every paradigm. The market mints a new “you’re already behind on this” every quarter. If you try to chase all of it, you’ll be exhausted and still behind. The engineers who last aren’t the ones who know the most tools. They’re the ones who can pick up whatever the moment requires, quickly, and be trusted to do it again next time.
Just-in-time learning
Think of it as just-in-time learning. You don’t stockpile every skill you might one day need, hoarding certifications out of fear. You pick things up right when the moment calls for them and trust that you can.
I learned the cost of getting this wrong, too. Toward the end of my time at Mint.com, we’d built something real, but we’d also become a bit of a dinosaur. Everyone around us had moved to test-driven development, git, and continuous integration and deployment, while we were still taking the site down in the middle of the night to push changes, costing engineers their sleep. Staying ahead isn’t about novelty for its own sake. It’s about not waking up one day to find the ground moved while you were heads-down.
But that’s only half the story. I’d also learned a ton in the four years I spent building that prototype, launching it, and scaling it to serve millions of users: how to work with cross-functional teams, how to iterate on a SaaS product, how to build trust with users around their financial data, and how to attract and retain them.
When I started my next startup, I took those skills with me, then filled in the technical gaps I’d missed: continuous integration and deployment, TDD, Git, managed hosting services, and NoSQL databases.
I kept adding to my toolkit over time. That’s the whole point.
It’s okay to be late
Now here’s the part that’ll feel illegal to say in 2026: it’s also okay to delay your learning.
I’ll be the first to admit I was not an early adopter of any of the chatbots. When they showed up in 2023 and 2024 and everyone lost their minds, I was in the thick of moving from Macs to iPhones at Apple — a whole new set of hardware technologies to absorb to support a new line of business. I also wasn’t sleeping through the night, because I’d just had my third kid. I did not have the bandwidth to bolt on one more thing, so I didn’t. In late 2025, when I had room in my schedule, I started working AI into my everyday life.
I want to be honest about why I could afford to wait: I had a steady paycheck the whole time. If you’re between roles right now, that particular luxury isn’t yours this month, which is exactly why the habit of staying lightly current matters most while you’re employed, when you have the cover to learn slowly. Build the muscle when it’s cheap, so you’re not trying to build it in a panic when it’s expensive.
But the deeper point holds for everyone: your brain needs recovery the way any overworked system does. Run it redlined indefinitely and the quality of everything degrades. Giving yourself permission to be a season late on the shiny new thing is not falling behind. It’s pacing yourself for a 20-year career instead of a six-month sprint.
If you’re reading this between roles
I need to say one thing directly, because the advice above can curdle if you’re hearing it freshly laid off.
Re-skilling is insurance you buy for yourself. It is not a penance you owe for getting laid off.
If there’s a gap on your resume right now, it is not a verdict on whether you kept up. Layoffs are business decisions dressed up as performance decisions, they happen to people who did everything right, who were promoted the year before, who were sharp the whole way through. “AI took my job” is real for some people and a flattering cover story for many more. So please don’t pick up the next skill as a way of apologizing for being let go. Pick it up because it gives you more options. That’s the only reason that holds up.
What to actually do this week
If you want to convert all of this into something concrete, here’s where I’d put the energy.
1. Pull ten real job posts.
For the role you genuinely want next, open ten current listings. Ignore the bloated wish-lists, and look for the skills that show up in eight or nine of the ten. That short list, usually just two or three things, is your real curriculum. Everything else is noise wearing a “required” badge.
2. Don’t blow your severance on a bootcamp.
When money’s tight, FOMO is expensive. A $15K program won’t save you, and it definitely won’t out-perform a few focused weekend hours on the two skills from your ten-posting list. Spend small and targeted, not big and scared.
3. Sharpen the meta-skill, not just the tool.
The durable skill was never memorizing the framework of the month, it was knowing how to find the answer fast.
In 2019, when I joined Apple, the plan was to shadow a coworker with a PhD and five years on the product. Two weeks in, he got pulled onto something else and I was in sink-or-swim mode. The habit that saved me wasn’t knowing the answers; it was getting good at figuring out how to find them, and leaning on my network when I was truly stuck, in ways that were worth their time too.
There were hard skills I learned and re-learned: how to program in Lua, Apple’s OS, and embedded systems. But there were soft skills that got me there too — like how to ask a question in a crowded Slack forum:
- Check whether anyone else has hit the same problem first.
- If not, share what’s different about your context: what you’re trying to do, what you’ve already tried, what worked, and where you’re stuck.
- Then wait patiently, but follow up, follow up, follow up, because you’re on the hook for delivering.
I also made it a point to attend internal training once a quarter. Companies need to provide breathing space for people to upskill, which is hard in a market that demands daily results. But while you’re employed, negotiate for the time you need to learn the new skill.
Getting good at figuring out how to find the answers, and discern the quality of those answers — is more valuable in the AI era, not less. AI is the most powerful “how do I find the answer” tool we’ve ever had. The people most afraid of being replaced by it are often the ones treating it as one more thing to memorize — instead of a thing to think with, and, of course, to judge the thinking of.
The takeaway
The market will keep inventing new things you’re supposedly already behind on. It did it with the web, with mobile, with the cloud, and now with AI. Every cycle, the same panic, a new mask.
You will not win by learning all of it. You’ll win by learning enough to stay ahead, proving you can pick up the rest fast, and resting enough to keep doing that for decades.
You’re allowed to be late. You’re just not allowed to stop moving.
If this resonated, tell me in the comments: what’s the one skill you’ll actually touch this week? I read every one.
