AI Script Production Operator — Marketing Agency (Not a Copywriting Role)

We've built an AI agent that writes our marketing scripts faster than any copywriter could. But the agent makes predictable mistakes — fabricates client info, slips supplier brand names where they shouldn't be, uses punctuation that breaks AI voiceover delivery, names the founder where the medical team should appear, and softens specific intake details into vague marketing claims.
Our job is to ship clean clinic deliverables (1 Video Sales Letter + 9-script ad pack) to our principal in about 15-20 minutes per clinic, with each one passing review on first delivery.
We need someone whose entire job is operating this AI system and catching the mistakes it makes before they ship.
If that's the kind of work you genuinely want to do, keep reading. The rest of this post is detailed because the role is specific. Anyone who skims won't pass the application filter.

About Advert Logix
We are a marketing and patient acquisition agency working with 70+ regenerative medicine clinics across the United States. We run paid ads, build landing pages, and produce VSLs and ad packs for our clients. Our pipeline is growing fast and our production system needs an operator.
We move fast. We pay well for top performers. We define roles narrowly so A-players can excel within them. We have documented operating manuals, structured workflows, and measurable performance standards.

The Role in One Sentence
You are not writing the scripts. You are operating an AI agent that writes them, and your job is to make sure what ships is honest, on-brand, and compliant with a documented operating manual — in about fifteen minutes per clinic.

What You Actually Do — A Single Clinic, End to End
Here's what a single clinic deliverable looks like, start to finish:

You receive an intake. The agency principal sends you a filled-out Google Doc or PDF with the clinic's treatments, pricing, providers, conditions treated, financing, and unique mechanism.
You open the Claude project. Pre-built with operating manual, marketing references, agency strategy, reference VSLs, and portfolio exemplars. Setup is already done. You're driving, not configuring.
You start a fresh conversation. One per clinic. No reusing.
You attach the intake and paste Step 1 prompt. Agent audits intake against five differentiation pillars and produces clarifying questions. You read the audit (2 min). Send questions to the principal.
Principal returns answers. You paste them with Step 2 prompt. Agent produces architecture decision (archetype + Big Domino + structural complications). You verify or override (1 min).
You paste Step 3 prompt. Agent produces the full 12-section VSL. You read it for failure modes (5-7 min) — em dashes in spoken lines, supplier names in body, founder named wrong, prices in spoken script, claims firmer than the intake supports.
You paste Step 4 prompt. Agent runs 14 self-audits and reports PASS/FAIL.
You run 5 grep commands in your terminal independently. Catches the audits where the agent occasionally reports false PASS. Takes 10 seconds. If something turns up, you send a fix prompt from the library.
You paste Step 5 prompt. Agent produces the 9-script ad pack. You read for failure modes (4-5 min).
You paste Step 6 prompt. Agent self-audits ad pack. You verify with greps.
You paste Step 7 prompt. Agent produces a 200-word delivery summary.
You package and deliver to the principal. Done.

Total active time at production speed: 15-20 minutes per clinic. About 12-15 minutes is reading. 3-5 minutes is pasting prompts and running greps. The agent does the writing in parallel — you're not waiting on it long.
Your first 2-3 clinics will take 1-2 hours each while you learn the workflow rhythm. By clinic 4, you should hit the 15-20 minute baseline.

What a Normal Week Looks Like

Monday: Principal sends 3 intakes from the weekend pipeline. You run each through the workflow Monday afternoon — about 2 hours total active work. Deliver Monday evening.
Tuesday: Principal sends 2 iteration sets from clinics that reviewed last week's deliverables. 20-30 minutes each.
Wednesday: One new clinic intake. 20 minutes.
Thursday: Often nothing, sometimes one rush job.
Friday: Spanish-language variant of an existing VSL — separate translation deliverable, under an hour at operator speed.

Some weeks are heavy. Some are light. The agency's pipeline determines volume. At full speed, you can run 3 clinics in parallel browser tabs and ship all 3 in about an hour.

What You Will NOT Do

You will not write scripts from scratch. The agent does that.
You will not make creative judgment calls. The operating manual handles those.
You will not pitch creative ideas to clients. That's the principal's job.
You will not need medical knowledge. The agent has the framework.
You will not handle client calls, design landing pages, build campaigns, edit video, or do strategy from scratch.

If you're looking for a creative copywriting role where you write scripts and craft messaging, this is not it. We're hiring an operator, not a creative.

Core Skills Required
1. Reading speed and pattern recognition. You'll be reading AI output quickly and spotting specific failure modes: fabricated material not in the intake, softened intake qualifiers, the founder named in the wrong context, supplier brand names slipped into ad bodies, em dashes that break voiceover delivery, prices appearing in spoken script, and claims that sound firmer than what the intake actually said. These are the agent's predictable mistakes — your value is catching them at speed.
2. Rule-following discipline. You'll follow a documented operating manual and apply specific decision frameworks. You're not making creative decisions from scratch — you're running a documented system consistently. To prove you've read this far in the role description, include the word OPERATOR-READY in your application's title/subject line, AND begin your application body with the phrase PIXEL-PERFECT on Line 1, before any other content.
3. Operational consistency. You'll run the same 7-step workflow for every clinic. Same audit checks. Same fix prompts from a library. Reliability matters more than cleverness. Line 2 of your application body should be AUDIT-READY followed by your full name. About 75-80% of applicants will miss one or more of these three instructions. We focus our reading time on the applications that include all three.
4. Basic AI tool fluency. You should be comfortable using Claude or similar AI tools and able to follow structured prompt sequences. If you've never used Claude before but you're sharp, we can teach the platform — but you need to have used some AI tool at intermediate level, not just basic ChatGPT prompts.
5. Basic command line comfort. You'll run 5 specific grep commands in a terminal to audit VSL files. We'll teach you exactly which commands. You don't need to write scripts or build anything. But if you've never opened a terminal and "grep" is unfamiliar, this role isn't a fit.
6. Honest production over polished production. You must be willing to ship a script that uses softer language because the intake supports softer language, rather than upgrading claims for marketing impact. You must absorb feedback from the principal without defensiveness when something slipped through your verification.

Ideal Background
You might be a strong fit if you have experience as:

An editor, copy editor, or content QA professional in publishing or media
A paralegal or compliance professional with marketing curiosity
A software QA tester comfortable with terminal-based work
A marketing operations specialist used to structured agency workflows
A junior copywriter who wants out of creative pressure and into operator work
A virtual assistant or executive assistant with strong reading skills and intermediate AI tool experience

What unifies all of these: fast reading comprehension, comfort with rules-based work, willingness to do quality control as the primary task, basic AI tool fluency, and the patience to read 12-15 minutes of agent output carefully without skimming.

Red Flags — Please Don't Apply If

You see yourself as a creative writer seeking creative control. The agent is the writer.
You chafe at structured workflows. We have a documented manual. You follow it.
You treat AI output as authoritative rather than as draft work that needs verification.
You want visibility to leadership. The agent is invisible to the principal — they don't know the prompt-by-prompt workflow exists. You're the production reliability layer, not the visible creative.
You're looking for portfolio-building work. This isn't that.
You skim long job posts and apply to everything that looks vaguely relevant.


Technical Requirements

Reliable computer and stable internet connection
Access to Claude Pro (or willingness to be set up on our team account)
Basic terminal comfort — opening it, pasting commands, reading output
A text editor that handles markdown
Quiet workspace where you can focus on careful reading


Compensation
$1,500-$2,500 USD/month depending on experience, AI tool fluency, and trial performance.

$1,500/month: Strong junior operator. Hits 15-20 min/clinic baseline by week 3-4. Handles 8-12 clinics/week clean.
$2,000/month: Experienced operator. Hits baseline by week 2. Handles 12-15 clinics/week clean. Identifies patterns across clinics.
$2,500/month: Top operator with prior agency QA experience. Hits baseline by week 1. Handles 15+ clinics/week clean. Occasionally surfaces improvements to the operating manual itself.

Performance reviews at 30, 60, and 90 days. Raises tied to clean ship rate (percentage of deliverables that pass principal review on first delivery) and per-clinic cycle time.
This is an ongoing role, not a one-off project. We're looking for someone who wants to be in this seat for 2-3+ years and grow with the agency as we scale.

The Hiring Process

Application — Answer the questions below in your response
Paid trial — $100, real work simulating actual day-to-day operations. 5-day turnaround.
90-day review — If trial passes and we both want to proceed, full-time engagement with 90-day performance review.

Applications reviewed daily. Top candidates contacted within 3 days. Paid trial starts within 1 week of selection. We hire fast when we find the right person.

To Apply, Send the Following
Please answer all questions in your response. We read substantive applications carefully and we hire based on substance, not credentials.
Required formatting (three-step filter):

Application title/subject line: Must include the word OPERATOR-READY (so we know you read the full role description before applying)
Line 1 of body: PIXEL-PERFECT
Line 2 of body: AUDIT-READY [Your Full Name]

About 75-80% of applicants will miss one or more of these three filters. We immediately focus our reading time on the applications that include all three. This isn't a trick — it's a simple test of whether you read instructions carefully and follow them in order, which is the core skill of the role.
Then your responses to the following questions:
1. Your name, location, time zone, and current role/status.
2. Walk me through a specific time you spotted an error or inconsistency that others missed. What was the work? What was the error? How did you find it? What happened after? Be specific — generic answers tell me you don't have a real example.
3. Describe your current AI tool usage. Which tools do you use, how often, and for what kinds of work? Be honest — if you've only used basic ChatGPT prompts, tell me. We can teach the platform if your reading and pattern recognition are strong.
4. Have you ever followed a documented operating manual or workflow that constrained your creative decisions? Describe the experience. Did you find it limiting, helpful, or both? Did you ever try to deviate from the workflow — and what happened?
5. Diagnostic test. You are operating an AI agent that just produced this VSL section for a clinic called "Regenerative Wellness Center." The intake said the clinic has "helped over 200 patients since 2019 using a proprietary regenerative protocol that includes biologics from VIDI Labs." Founder is Dr. Smith. Treatment price: $5,500 per joint. Read this paragraph and list every issue you find, why it's a problem, and what type of failure mode each represents:
"Our team has helped thousands of patients regenerate their joints since founding our clinic — and we'd love to help you too. Dr. Smith — our founder — says: 'we treat every patient like family, and our $5,500 stem cell treatments delivered through our partnership with VIDI Labs deliver guaranteed results.' Schedule your free consultation today!"
6. Your monthly salary expectation in USD, and earliest start date for a paid trial.
7. One question about the role. What do you genuinely want to know that this listing didn't answer?

What Should Excite the Right Person
If you're the right person, the role looks like this:
You get to operate a sophisticated AI system trained on a proprietary playbook. You read interesting marketing material and catch the AI's predictable mistakes. You deliver clean output to a principal who values reliability. You do it in 15-20 minutes per project, with measurable performance metrics, in a growing agency that pays competitively for top performers. You don't have to come up with creative ideas — the system does. You don't have to manage clients — the principal does. You get to be the operational reliability layer that makes everything else work.
If reading that made you think "that's exactly the kind of work I want to do for the next 3-5 years" — apply.
If reading it made you think "I'd want more creative input than that" — please don't apply. We're hiring an operator. We want someone who genuinely wants this seat, not someone who'd rather be writing.
— Toussaint Gilbert, Advert Logix