How I actually work
"Read the surface. Pick the worst-pulling part. Rewrite one specific thing and tell you why. No three-option menus. No 40-slide audit deck."
See the full sample→How I actually work
"Read the surface. Pick the worst-pulling part. Rewrite one specific thing and tell you why. No three-option menus. No 40-slide audit deck."
See the full sample→Free sample
Want a taste of the operator stack first? Mini-Atlas: 3 prompts I paste in my own work. Plain text. No email gate.
Read Mini-Atlas→One idea into seven posts. Voice-loading workflow that stops AI from sounding like AI. The fix is not better prompts.
You opened ChatGPT three times this month and closed it without finishing. This is the 5-prompt entry point.
You ship alone. No one reads it before it goes out. Then 24 hours later you spot the thing a second pair of eyes would have caught in 10 seconds. This is the pre-ship pressure test I built so I would stop having that morning.
You are doing ops, content, and sales by yourself. The bottleneck is decisions, not effort. This is the kit for that.
You spent a weekend rehearsing STAR stories and got blindsided in minute four. The candidate who prepares the right 10 questions beats the candidate who prepares the wrong 50.
More kits drop daily this week
No spam. Unsubscribe in one click.
You are a senior IC or a director. Your hour is worth more than the price of this kit, which means the DIY version is the wrong math for you. This is the one where I do the work. Not a junior, not a $5 freelance pass-through. Me, reading...
47 applications, two callbacks. Two filters kill resumes before a human reads them: the ATS parser and the 6-second scan. This kit fixes both in one pass.
Most SOP libraries document the happy path. This one documents what breaks and how you recover.
The email says "we would like to extend an offer," and you freeze. Not because you do not want more. Because you do not have the words. The freeze is a script problem, and the first 24 hours after the offer set the floor for the next sev...
An AI agent with write access escalates privilege on its first hallucination unless something constitutionally refuses the bad writes.
Audit-to-close rate is low. Three exact leaks beats forty generic ones. Always. This is the 30-minute teardown that earns the reply.
One cold template does not fit one list. Three shapes plus the 8 question rubric that picks the right one in 90 seconds.
You've audited the same pattern three times because nobody wrote down what got decided last time. Discipline travels better than artifacts.
Forty tabs, twelve docs, three spreadsheets, one notebook. The job hunt is not hard work, it is lookup cost. This is the consolidation.
The four-asset sequence that earns the first audit. For the $150 to $400 an hour solo consultant who hit a referral ceiling.
Launch math is still vibes. This is the seven-axis rubric with a threshold table that says ship, fix, or kill.
You either run review loops on decisions that should ship solo, or you ship decisions solo that should have had five voices on them. Both fail the same way.
You hit publish. Replies arrive. You miss most of them. The next post is shaped by whichever reply was loudest, not the one that mattered.
Peers who grew up here got the rules at the dinner table. You didn't. This is the manual, not the motivation.
Mid-career, AI moving fast, the corporate HR ladder feels narrower every year. The 30 day pivot out of HR into operator work.
Generic STAR coaching is everywhere. But once you can see the questions the recruiter will actually ask, generic prep is over. This is the 12-question generator.
You're converting under one percent and you can't tell which slot is bleeding. Landing pages rarely fail on the hero alone. This is the seven-slot diagnostic that finds the leak.
You got the meeting invite, the severance packet, and the access-revoked email in the same week. Layoff drag is rarely a money problem in week 1; it's a decision-order problem.
Two agents touched the same file in the same minute. One overwrote the other and nobody noticed for an hour. This is the rule that stopped it.
Your newsletter sounds like a brand-voice committee wrote it. The drift is invisible to you because it happens slowly, one issue at a time, until the thing reading back to you isn't yours anymore.
Your AI drafts read like every other AI draft. Same em-dashes. Same credential-recital openers. Same fabricated kitchen-table stories.
You want a fast read on whether AI fits in a workflow before you spend a dollar. Most AI audits ask 50 questions and answer none. This is one page.
You read $100M Offers and the frame did not map to your solo practice. That's not your fault. Public frames optimize for the median reader, and a solo operator is not the median.
You ran the 5 starter prompts. They worked, and now you want a deeper sampler. One prompt per work mode, not five flavors of the same one. This is the bridge to the Atlas.
The algorithm doesn't punish AI content. The audience does. This is the 90-second gate before you post.
Cold or warm outreach and your opener keeps falling flat. Generic openers underperform reference-specific openers in real reply data.
Written from a CSV, not theory. 1,200 sends, 47 replies, 11 calls, 4 clients. Reply-first templates plus the rubric that picks the right shape.
Dozens of applications out. A handful of replies back. At some point the math stops being a numbers game and starts being a resume problem. This is the single-prompt rewrite to get unstuck, free.
Rewriting a resume is the easy part. Knowing when the rewrite is actually done is where most people stall. This adds the one thing the Lite leaves out: the scoring rubric that tells you.
A prospect replies STOP. You miss it. The next send goes through. The complaint lands at your ESP. Your account gets suspended. The cost is recoverable; the relationship is not.
Got laid off in the last 8 weeks. Standalone tools are easy. Sequencing them is the hard part. This is the four-day plan.
You keep opening ChatGPT and closing it without finishing. Casual AI use never compounds. Named workflows do. The 24 hour install.
A Notion workspace for the operator using AI daily. Planner, tracker, prompt library, and the small set of templates that compound.
You run more than one brand. After a month, the voices start blending. Brand A's palette shows up in Brand B's deck. Brand C's voice creeps into Brand D's newsletter.
You posted for three weeks, got eight likes, and started questioning the path. Most consulting content optimizes the visible parts and skips the load-bearing ones. This is the boring middle.
I almost deleted a folder of family photos because it was named like a duplicates pile. It held files that exist nowhere else. That near-miss is why every prompt in this pack carries one hard guardrail: the AI is never allowed to suggest...
Forty businesses. All dead. The patterns that killed them are the patterns the survivors warn you about, but only after they've survived.
Your AI agent reverts to its training default after three conversations. Helpful, cautious, mid-Atlantic, indistinguishable from every other assistant.
You publish weekly. Your follower count grows. Your impressions look fine. Revenue doesn't move. The leading indicator you're tracking is the wrong one.
Auditors keep re-flagging the same code as a bug. Most repos have no record of which patterns the team already decided are deliberate.
Code reviews check style. The four principles check whether the code earns the merge. Most ship-blocking bugs slip past style checks.
Three AI agents, and they've started stepping on each other. Most multi-agent content is a launch announcement. This is the maintenance manual.
Most 12-week years drift in week five. Three vital goals collapse into one anyway and the other two lanes were just calendar noise.
You keep rewriting the same prompt from scratch. A survivor list, not a wishlist. 25 prompts pasted more than twice.
Founder is the bottleneck and the only one who can fix it. $79 covers the workflows. $147 covers what the owner does that the employee cannot.
You set up an AI workflow, used it for a week, then forgot to notice when it stopped working. AI stacks decay silently. This is the 20-minute Sunday check.
Two sessions on one tree both commit. Branch HEAD flips. Work disappears. The post-mortem says it was your fault but you don't see how.