On August 26 a paid subscriber sent me a DM.
He had taken the Mercor profile prompt I send to paid subscribers, rewritten his profile to target the Generalist track, and hit save. Within minutes he had an offer in his inbox. No interview. No waitlist. No “we will get back to you.”
Two days later he wrote again, and this is the half most newsletters would cut. The role was not what he expected. He was thinking about pausing, and he wanted to know whether re-targeting his profile a second time would get his account flagged.
Getting the offer turned out to be the easy part.
That gap, between landing the work and keeping it, is the whole reason I am opening the paid tier up properly this week. But first, here is the actual thing he changed, free, because I would rather show you the work than describe it.
Where this newsletter came from
Four months on Mercor, thirty seven thousand dollars, and no idea at the start whether any of it was real.
I wrote that up in How I made $37k in 4 months on Mercor and the response is what created World Model.
How I made $37k in 4 months on Mercor (And You Can Too)
There is a new category of jobs quietly emerging in the tech world. It doesn’t require you to live in Silicon Valley, and it doesn’t require a traditional corporate resume.
My inbox filled with people asking the same thing: not “what is this,” but “how do I actually survive it.” So I wrote the Hands-On Mercor Playbook, and then I kept going.
Hands-On Mercor Playbook
A few weeks ago, I published a breakdown of how I made $37,000 on Mercor over four months. The response blew up. My inbox flooded. People wanted more—not just the “what,” but the “how.” How do you actually survive and thrive on a platform where projects vanish overnight, the rules change mid-game, and your “employer” is technically a language model that hasn’t learned to think yet?
I started this because I could not find a jobs list I trusted. Every roundup was copy-pasted from aggregators, half of it stale, some of it never real. So I built the opposite: I apply to the listings myself, from my own account, and I keep the receipts. When I cannot confirm something, I say so in the post. In the last free issue I applied to twelve Mercor tracks by hand across three days and told you which three turned me down.
Twenty three thousand of you now read this, from every timezone I can name. World Model is currently rising in Substack’s Technology category. I still do the checking by hand.
The part I have to be straight with you about
You have noticed the shape of the deal.
The platforms train AI. The platforms also, on most projects, prohibit you from using AI to do the work. And then they set a maximum handling time that assumes a pace a careful human can barely hold. Go too slow and you are flagged for efficiency. Go too fast and you are flagged for quality. Nobody tells you where either wall is. I mapped what I could in The AHT Game.
The AHT Game: How Slow Can You Be Before Mercor Cuts You
Last episode I wrote that every Mercor project has a corridor of acceptable speed, that nobody tells you where its walls are, and that people get terminated for hitting either one. Dozens of you wrote back with the same question.
That is a real bind and I am not going to pretend otherwise.
I am also not going to sell you a way around it. Accusations of LLM use are one of the live termination reasons on these platforms right now, and termination usually means the work you already finished does not get paid. Advice that gets you offboarded is not a shortcut. It is the most expensive thing you can buy.
So the World Model position on speed is narrow, unglamorous, and it holds up:
Speed comes from what you do before the timer starts.
Know the rubric cold before you open task one.
Build your reference sheets, templates, and answer patterns between sessions, on your own time, where no project policy applies.
Learn the corridor on your specific project, because the cap and the tolerance are both set per project, not per platform.
Find out what tooling your project actually permits. It varies more than people assume, and the answer is usually written down somewhere nobody reads.
Recognize when a cap is unwinnable and decline the project. This is the highest-value decision on the list and almost nobody makes it.
I updated my resume and got an instant Mercer offer
Last week I logged into Mercor, made some edits to my profile, hit save, and within minutes got an offer in my inbox. No interview. No waiting list. No “we’ll get back to you.” Just an offer, ready for me to accept.
None of that is a trick. It is just the difference between contractors who last nine months and contractors who last three weeks.
What the paid tier actually is
It is this newsletter’s verification work, done for one profile instead of a public list. Yours.
You tell me where you are. I go find what fits, and I tell you what I would do if I were you.
1. The profile prompts. The Mercor one that produced the offer above, plus versions for micro1, Handshake AI, and the smaller platforms. Updated when the platforms change, which they do.
2. Matched openings, sent to you. AI training and expert-data roles across the big platforms and the small ones, filtered against your background, your region, and your license. Including postings that never surface on aggregators because they close before anyone indexes them.
3. Remote work operations. Running a distributed week, handling standups across a nine hour gap, managing Slack so it does not manage you, tracking your own hours independently so you can dispute a shortfall in the same week it happens.
4. The offboarding-risk playbook. Where the handling time corridors sit on projects I have run, what the assessment gates actually test, and how to exit a project without it costing you the next one. Built from more than ten projects on both the contributor and the reviewer side.
I cover Mercor, Handshake AI, and micro1 in the same place, which matters more than it sounds like. They are not substitutes. Mercor pays higher at the top and goes quiet for weeks at a time. Handshake is steadier and lower, and has its own payment problems that I have documented and will keep documenting. Holding both queues open is how a bad month on one stops being a zero month.
How to start, in one reply
Upgrade, then reply to the welcome email with three lines:
Which platform or platforms you are on.
Which stage you are at: not applied, applied and waiting, onboarded and idle, or actively on a project.
Your background, your region, and any license or specialty.
That is enough for me to start matching. You get the profile prompt back the same week, and your first matched roles in the next customized email.
What this is not
I do not sell jobs, and I have no back door into any platform. Nobody does, and anyone telling you otherwise is selling you something worse than a subscription.
What I have is a list of roles I checked myself, several years of contractor scar tissue across both sides of these projects, and enough hours in the queue to know which silences mean something and which ones do not.
The subscriber at the top of this post got an instant offer and then found out the role was wrong for him. He is not a testimonial. He is the accurate version of this work: the door opens faster than you expect, and then the real question starts.
I would rather be in your inbox for that second part.