A Pocket Guide to Surviving the Robot Apocalypse
February 16, 2026
Everybody’s talking about the robots taking our jobs. The CEO of Microsoft AI thinks white-collar workers have “12 to 18 months”. The CEO of Anthropic, a frontier LLM provider, predicts the end of engineering on a similar timeline. Engineers have felt the disruption first, and Matt Shumer captured that feeling accurately in Something Big is Happening.
12-18 months is likely overselling the claim—I expect the transition will be a lot slower than that in reality, and hit different job functions at different times, perhaps years apart. And there’s still some chance that AI progress tops out at sub-human levels (although that is becoming less probable over time).
At a minimum, I would like you to entertain the possibility that an AI will be able to do your job at least as well as you in the next few years. And if you believe that’s possible, you should spend some time thinking about the right way to react.
Weirdly, I’ve been thinking about this for a long time. In 2023 my co-founder and I launched Taxy, the first working AI-powered browser agent. A year prior I built Clippy AI, the first full-file AI-powered code editor plugin. And of course there’s OpenPipe, the AI training startup I co-founded in 2023 and sold last year.
Over the last two months it has been surreal to see both (a) the concrete evidence that the full automation of software development is no longer theoretical, but visibly approaching quickly, and (b) the wider industry waking up to this in a bit of a panic. Many friends and co-workers grappling with this new reality have reached out wanting to swap notes on what this means for them.
I’d like to share what I’ve done to prepare for the oncoming turn. I’ve broken this down into three simple pieces of advice, learn, earn and return, that I hope will be helpful to any knowledge worker who, facing the possibility of future automation, isn’t sure how to react.
“At dawn, when you have trouble getting out of bed, tell yourself: ‘I have to go to work — as a human being.’”
— Marcus Aurelius
Advice #1: Learn
Your first task is to learn how the current frontier AI assistants work. Not the mostly-useless free tiers. Not the model you tried out 18 months ago that hallucinated citations. Get on the paid plan of a frontier AI provider and immediately switch to the strongest model they offer. Going forward I’ll use “Claude” generically in this article, but ChatGPT is about as good for everything I mention.1 Ask it harder and harder questions about your area of work until you stump it.
If you really want to see what the latest models can do, you’ll also need to give them access to your personal life—email, text messages, documents, etc. It’s incredible how much more helpful they can be when they understand you better. Unfortunately this still isn’t super easy to get set up, so I’ll talk more about how to do this in a follow-up post.
To make things more concrete, here are a few workflows I’ve used Claude to either partially or completely automate over the last month:
- Automated used-car search over Craigslist, FB marketplace, and local sites (I have a specific set of requirements that normal filters don’t capture).
- Plan a multi-country family vacation to visit my wife’s parents in Europe.
- Review bloodwork results and suggest follow-up tests and nutrition/lifestyle changes.
- Compute my estimated taxes and prepare a draft return for review by a CPA.
If you’re not the sort of person who tinkers with new tech for fun, this may sound like work. And it is. But do it anyway. This is the best short-term investment in your career you can make in 2026. Even if the robots do come for all our jobs, being known as the person who uses AI best will likely push your relevance months or years further.
Advice #2: Earn
We are entering an era of extreme disruption. The future is radically uncertain. The people building frontier AI models aren’t sure what their long-term impacts will be, and you can bet that the Washington think tanks and labor market economists have no clue.
In uncertain times you should spend less and save more. If you’re living paycheck to paycheck, start building towards a 3-month emergency fund. If you already have that, 12-36 months of savings wouldn’t hurt either2.
By the way, this is another fantastic place to put advice #1 in practice. Dump your last 3 months of credit card statements, bills, and pay stubs into Claude/ChatGPT. Ask it to analyze your spend across categories and suggest areas you could optimize. Tell it to identify subscriptions you might not be using, or vendors that have cheaper alternatives. I’ve done this several times over the last year and gotten highly actionable advice that has saved me literally thousands of dollars with very little effort.
Diversify across both standard financial assets and less traditional ones. Again, ask Claude for advice, given your existing portfolio, long-term goals, and specific risk tolerance. I’ve used financial advisors in the past and Claude is better than 95% of them. (That said, don’t invest in an asset class that you don’t understand or trust, since you won’t have the conviction to hold through volatility. And we’ll likely see a lot of volatility in the short term.)
Advice #3: Return
Now comes the real talk. The time may come (and I hope it doesn’t!) where even if you do (1) and (2) above, your industry fundamentally changes in a way that drastically decreases the number of jobs available. This has happened many times in the past, and will almost certainly happen again.
To complicate things further, every technological revolution (electrification, PCs, internet, mobile) has propagated faster than the last, and this one is on track to continue that trend. We’ll likely see it play out in years, not decades.
The development and deployment of thinking machines—machines that, in the scenario we’re considering here, think better than us—will radically disrupt existing economic, social, and political structures. This is scary! Even if we end up in a better place on the other side, that disruption will have a cost that may show up in unexpected places.
It’s natural to feel a bit anxious, given that reality. But let me share what has most helped me, as I’ve had to come to terms with the future’s growing uncertainty. Perhaps ironically, the best mental framework for dealing with this uncertain future is a return to certain traditional principles from the past.
Like many of you, I came of age in a stable, developed country. I grew up assuming that stability is the default. From a wider perspective though, it emphatically isn’t. Living with the prospect of mass disruption, and perhaps even existential risk, is, sadly, a return to the historical norm. A drought, a plague, a generational grudge match with the next city over—these things have always threatened human civilization, and the few decades we’ve enjoyed without them are very much the exception to the rule.
Our ancestors coped with this by finding meaning beyond the chaos; even in the chaos. We can too.
What does this mean concretely? Here are some suggestions that help me stay grounded and optimistic:
- Stay connected to your basic human needs. If you aren’t exercising, set a goal to walk 10k steps a day. If you aren’t getting enough sleep, commit to turn off screens by 8pm. If you don’t have close personal connections, join a book or running club. Life is much happier when these fundamental human needs are met. At the risk of sounding like a product pitch, this is again a place where Claude can help—ask it to find local clubs specific to your interests, or to make an exercise plan tailored to your goals and capabilities.
- If life still feels difficult, consider the philosophy of the Stoics. Finding and living the good life in spite of the challenges the world threw at them (and it threw a lot!) is the core of the Stoic philosophy. If you’d like a book recommendation I love Marcus Aurelius’s Meditations, which has given me both an ideal to strive for and great peace in troubled times.
- If you’re a religious person, consider what your religion says about hard times. It was written by those who faced them.
And Yet…
I’ve focused on the negatives here since you have to start by protecting your downside. But it’s not all doom and gloom. In fact, I believe the most likely outcome is that 10 years from now we’re in a world that is far better for most humans. I love Nick Bostrom’s book Deep Utopia—it gives concrete examples of ways we can find joy and meaning in a world where we don’t have to do anything to survive.
I believe that world is achievable. The transition will inevitably be rough, but the destination is a worthwhile one if we can reach it.
“Here is a rule to remember in future, when anything tempts you to feel bitter: not ‘This is a misfortune,’ but ‘To bear this worthily is good fortune.’”
— Marcus Aurelius
I want to deeply thank Matt Shumer, Saumya Gandhi, Karla Inostroza, and Karen, David and Scott Corbitt for their review of earlier versions of this essay.
Footnotes
-
I personally use a mix of all three major providers depending on the task. OpenAI’s Codex with GPT 5.3 handles most of my deep-thinking coding at work. Claude Code with Opus 4.6 is my go-to when I need web browsing automation, since it connects to Claude for Chrome. And Gemini gets a lot of my research work or anything on the go — it works really well with Android and has the best web search of the three. All that said, if you want to stick with one ecosystem, I’d recommend Claude for now. Opus 4.6 is a great all-around model, in first or second place for everything I do. ↩
-
On this, I put my money where my mouth is. My co-founder and I sold our startup in 2025 because I felt this change on the horizon and wanted to diversify my assets. It was an incredibly tough call; the company was in a strong place and I loved the work. The natural next step was to raise another round and triple-down. But selling was the right decision for us. I sleep a lot better at night with a less concentrated financial position. ↩