In December, I canceled a $3,200/month content retainer.

Not because the agency was bad. Because I built a 4-prompt workflow that produces the same output — in 90 minutes a week instead of waiting 5 business days for a draft.

This issue is that workflow. The full thing. Every prompt, in order, ready to run.

If you pay for a content agency, a copywriter retainer, or a marketing assistant — read this carefully. Because what I’m about to hand you is the operating system they’ve been running behind the scenes, minus the markup.

What Agencies Actually Do (And Why AI Does It Cheaper)

Agencies and retainer writers are not magic. Their process — stripped of the client calls, the Slack threads, and the account manager — looks like this:

Get a brief or a topic

Research the subject

Write a first draft

Edit for tone and audience

Repurpose across formats

That’s it. Five steps. Each one is a prompt.

The reason agencies charge $2,000–$5,000/month for this is not because the process is complex. It’s because coordination, revisions, and communication take time — and time has overhead.

When you become the operator and AI becomes the system, the overhead disappears. The output doesn’t.

What stays yours: the strategy, the topics, the editorial judgment, and the final approval. What AI handles: the execution.

This is not the future. Operators are running this system right now. Some of them are charging clients for the output and quietly keeping the margin.

The Content Engine — 4 Prompts in Sequence

This workflow produces a complete piece of content: researched, drafted, edited, and repurposed for every platform. Total active time: 30–45 minutes. Total elapsed time: 90 minutes including review.

Run these prompts in order. Each one feeds into the next.

PROMPT 1 — The Research Brief

Before you write a single word, you need a foundation. This prompt builds it.

Role: You are a senior content researcher who specializes in creating research briefs for professional writers. You find angles, data points, and perspectives that make content stand out from generic coverage.

Context: I’m writing a piece of content about [TOPIC]. My audience is [DESCRIBE AUDIENCE — e.g., “mid-level marketing professionals at B2B companies” or “college students starting their first job search”]. The goal of the content is to [SPECIFIC GOAL — e.g., “establish credibility on this topic” / “drive sign-ups to a free resource” / “answer the question my audience is already Googling”].

Task: Create a research brief for this topic that includes: 1. The most surprising or counterintuitive angle I could take (the one that makes someone stop scrolling) 2. 3–5 specific statistics, studies, or data points I should reference (with enough detail to verify them) 3. The 3 most common takes on this topic — so I can differentiate from them 4. The strongest opposing argument to my intended angle 5. 5 specific sub-questions my audience probably has about this topic 6. A recommended headline structure (not the final headline — just the type: list, how-to, question, provocative claim, etc.)

PROMPT 2 — The First Draft Engine

Feed the research brief from Prompt 1 directly into this one.

Role: You are an expert content writer who specializes in [YOUR CONTENT FORMAT — e.g., “newsletter issues”, “LinkedIn articles”, “blog posts”]. Your writing is known for being direct, specific, and useful — never vague, never padded.

Context: I have a research brief for a piece about [TOPIC]. My audience is [AUDIENCE DESCRIPTION]. The piece should run approximately [TARGET LENGTH — e.g., “800 words”, “1,200 words”]. My brand voice is [DESCRIBE VOICE — e.g., “direct and no-nonsense, like a smart colleague talking to you over coffee, not a consultant presenting to a boardroom”].

Here is my research brief: [PASTE OUTPUT FROM PROMPT 1]

Task: Write a complete first draft of this content. Specifically:

  • Open with a hook that would stop someone mid-scroll (a surprising fact, a specific scenario, a direct statement that challenges an assumption)

  • - Do not start with “In today’s world” or any variation of that phrase

  • - Make every paragraph earn its place — if a paragraph doesn’t advance the argument or add value, cut it

  • - Include at least 2 of the data points from the research brief

  • - Close with a specific, action-oriented takeaway — not a vague “start your AI journey today” type of line

Format: Write the full draft. No commentary before or after. Just the content.

What you get: A first draft in about 90 seconds that incorporates your research, matches your voice, and starts with something other than a cliché opener. It won’t be perfect. That’s what Prompt 3 is for.

PROMPT 3 — The Editor Pass

Run this on the draft from Prompt 2. Do not skip this step. The first draft is always fixable.

Role: You are a demanding but constructive content editor who has worked with top-tier publications. You have zero patience for filler, hedged language, or generic phrasing — and you’re equally skilled at fixing what you flag.

Context: I have a content draft that needs editing before publication. My target audience is [AUDIENCE]. The goal of the content is [GOAL]. The intended platform is [WHERE IT WILL PUBLISH].

Task: Edit the draft below for the following:

  1. Cut every sentence that isn’t doing work (flag it and explain why)

  2. 2. Flag any claim that needs a source or sounds unsupported

  3. 3. Identify the weakest paragraph — rewrite it

  4. 4. Check the opening: does it earn the read within the first 3 sentences? If not, rewrite it.

  5. 5. Check the close: does it end with something the reader can do or think immediately? If not, rewrite it.

  6. 6. Flag any phrases that sound like “AI wrote this” and replace them

Format:

— First, show me the edited version in full

— Then show me a “changes made” list (what you cut, changed, or rewrote and why)

[PASTE DRAFT FROM PROMPT 2]

What you get: A cleaned-up draft with a record of what changed and why. The “AI-ish phrase” check alone is worth running on every piece of content you publish.

PROMPT 4 — The Repurposing Engine

One piece of content. Five platforms. Ten minutes.

Role: You are a social media content strategist who specializes in repurposing long-form content into high-performing short-form posts for multiple platforms.

Context: I have a finished piece of content about [TOPIC]. My brand voice is [VOICE DESCRIPTION]. I publish on [LIST YOUR PLATFORMS — e.g., “LinkedIn, X/Twitter, Instagram”]. My audience on each platform: [BRIEF DESCRIPTION PER PLATFORM IF DIFFERENT].

Task: Repurpose the content below into the following formats:

  1. LINKEDIN POST (600–900 words)

  2. - Open with a hook line (no “I’m excited to share” — open with the idea)

  3. - Use short paragraphs (1–2 sentences each)

  4. - Include one data point or specific example

  5. - End with a question to drive comments

  1. X/TWITTER THREAD (6–8 tweets)

  2. - Tweet 1: The most provocative or surprising idea from the content

  3. - Tweets 2–6: One insight or step per tweet

  4. - Tweet 7: CTA to subscribe/read more

  5. - Keep each tweet under 280 characters

  1. INSTAGRAM CAROUSEL (7 slides — copy only)

  2. - Slide 1: Hook headline (5 words max)

  3. - Slides 2–6: One insight per slide, 1–2 sentences each

  4. - Slide 7: CTA

  1. SHORT-FORM REEL/TIKTOK SCRIPT (60 seconds)

  2. - 0–5s: Hook (state the most counterintuitive insight)

  3. - 5–45s: Walk through the core system step-by-step (speak as if talking to one person)

  4. - 45–55s: State the result

  5. - 55–60s: CTA to subscribe

Format: Number and label each format clearly. Write each one completely — no placeholders.

[PASTE EDITED DRAFT FROM PROMPT 3]

What you get: A full week of social content from a single piece of writing. This is the workflow that lets a single operator run omnipresent social without a social media team.

The Math on This

Before the operator model: 1 content piece per week, outsourced for $3,200/month, delivered in 5 business days, 2 revision rounds, one platform.

After the operator model: 3–4 content pieces per week, produced in 90 minutes each, delivered in real time, unlimited revisions, five platforms.

The agency was not expensive because content is hard. It was expensive because the coordination of humans is hard.

AI eliminated the coordination layer. The content quality — with the right prompts — is comparable. And in some dimensions it’s better, because AI doesn’t have conflicting priorities, a vacation calendar, or a creative block.

This is not an anti-agency argument. Agencies offer things AI cannot: relationships, creative strategy, institutional knowledge, accountability. For certain work at certain budgets, they are worth every dollar.

But for execution-layer content — the drafts, the repurposing, the formatting, the platform-specific rewrites — the operator model runs circles around the retainer model in cost, speed, and volume.

Who Needs This Right Now

Solopreneurs and freelancers: You’ve been trading time for content. This frees that time for the work that actually grows your business.

Small business owners: You’ve been paying for the coordination layer. Now you just pay for the strategy.

Marketing professionals: This is the workflow your agency is running. Knowing it makes you a better client — and might make you an operator worth twice your current salary.

Students and early-career professionals: Learning this system now puts you in the top 3% of AI operators before you’ve written your first real resume. The people hiring in 2026 are looking for exactly this.

Every week, Nova AI delivers one system like this — a complete, deployable workflow with every prompt written out and every step explained.

If this issue saved you two hours — or $800 of agency spend — the archive is full of more.

And if you want the complete 20-prompt Operator Pack (every prompt organized by category, RCTF-formatted, pre-built for the 20 tasks operators face most) it’s $17 on Gumroad. That’s less than an hour of the retainer you’re about to cancel.

→ The Operator Pack — novamedia42.gumroad.com

— Nova AI

The system for operators. Free, weekly, no fluff.

P.S. The single most common question we get is: “Which AI model should I use for this?” Next issue: we tested Claude, ChatGPT, and Gemini on 12 real work tasks. The results were not what we expected — and not what the marketing teams at those companies would want you to see.

Nova AI | novaai.media | 617 Vista San Javier, San Diego CA

Format: Use headers for each section. Be specific — I need material I can build from, not general suggestions.

Keep Reading