Laid Off From Big Tech? Here's How to Build Your Own Thing
A 90-day plan to validate and build your own thing instead of waiting for the next job.
Here's a number worth sitting with: 56% of the tech layoffs announced in 2026 explicitly name AI, automation, or machine learning as the reason, and between them they account for 164,971 people across 150 companies, per Crunchbase's tech layoffs tracker. That's not a soft patch in the market. That's a structural call made somewhere you weren't in the room, using a tool you've probably got open in another tab right now.
If you're one of the 164,971, the standard advice is "network harder, get another job." It's not wrong exactly, it's just slower than most people admit, because the applicant pool for every open role now includes everyone else who got the same email you did. The faster path, for a specific kind of person reading this, is the one you're already halfway suited to: build your own thing, solo, and find out fast whether anyone else wants it.
Here's the shape of the next 90 days.
Find the idea inside your last job, not outside it
You don't need a brainstorm. You need the process your team quietly patched over for two years, the spreadsheet you built because the \"real\" tool didn't do the one thing you needed, the complaint that came up on every customer call and never made the roadmap. The problems you're best positioned to solve are the ones you were closest to, which means the idea search is mostly a memory exercise, not a creativity one.
Prove it before you build it
A landing page and a waitlist tells you whether the way you describe the problem stops anyone scrolling. Ten actual customer conversations (not a Typeform, an actual conversation) tell you whether the problem is as sharp for other people as it was for you. A paid pre-order tells you the one thing enthusiasm never will: whether someone hands over a card number before the thing exists. If nobody will pay for the promise, you've saved yourself three months of building the wrong thing.
Write the go-to-market plan in an afternoon
Three questions, answered specifically, not aspirationally: who are your first ten customers (name them, or describe them so precisely you could pick them out of a room); where exactly do you find them, one channel, not five; and what do you actually say to them, that isn't \"check out my new thing.\" If the plan takes longer than an afternoon to write, it's a plan for a company with a lot more runway than you currently have.
Build the smallest version that proves the point, then stop
No-code handles the interface. Claude and Cowork can run and build a surprising chunk of the actual system, not just help you word the landing page. Cursor takes the code you do need to touch yourself. None of it replaces validation, it just shrinks the distance between "I have an idea" and "a stranger can use this" down to something you can cover solo, before you'd have even finished onboarding at your old job. The moment to bring in help is when you're duplicating a skill you genuinely don't have, not before you've proven a single person wants the thing.
Some of what you build in the next 90 days won't work. That's normal, and it's a much cheaper way to find out than spending those same 90 days waiting for a recruiter to call back. You didn't choose this reset. You do get to choose what you build with it.
Sources and citable claims
56% of tech layoffs announced in 2026 explicitly cite AI, automation, or machine learning as the cause, affecting 164,971 people across 150 companies.
Source: Crunchbase tech layoffs tracker, 2026.
Find the first people most likely to want what you built.
Give Romy a URL or one line about your product. Get a first read on who to focus on, where to test the market, and the angle worth trying first.
