--- skill: Article Accelerator version: 1.0 source: The Mantle Institute — mantleinstitute.com/playbook/article-accelerator layer: school sfia_minimum: 3 model_maturity: L3 tags: [content, writing, knowledge-work, adversarial-review] license: CC BY 4.0 --- # Article Accelerator A human-led, AI-assisted workflow for producing rigorous, adversarially-reviewed written content — from rough idea to publish-ready article — at a fraction of the usual time investment. The human drives direction, argument, and voice throughout. The AI carries the drafting weight, surfaces blind spots, and challenges assumptions before anything goes public. This is not AI writing for you. It is AI working alongside you, at pace, while you retain authorship of every idea. --- ## How to use this skill Load this file as context at the start of a session. Tell your AI: "We are working through the Article Accelerator workflow. Follow the steps in order. Do not proceed to the next step until I confirm I am ready." This keeps the AI in a supporting role rather than driving ahead of your judgment. --- ## Workflow ### Step 1 — Originate the argument Before touching AI, articulate the core claim in one sentence. What do you believe that most people in your field have wrong, or have not yet said clearly? If you cannot state it, the article is not ready to draft. ### Step 2 — Brief the model Share your one-sentence claim, your intended audience, your approximate length, and any evidence or examples you already have in mind. Include your writing constraints: voice, platform, tone. ### Step 3 — Generate a structure Ask the AI to propose a structure: three to five sections, each with a one-line purpose. Review it. Push back on anything that buries the argument or decorates rather than advances. ### Step 4 — Draft section by section Work through the structure iteratively. Do not ask for a full draft in one pass. Each section should be reviewed and steered before moving forward. This is where your judgment shapes the output. ### Step 5 — Adversarial review (three cycles) Once a full draft exists, run it through three cycles of structured critique using at least two alternative models — ideally from different vendors than the one that drafted the article. Each cycle uses a different persona: - Cycle 1: A sceptical senior practitioner in the field - Cycle 2: A first-time reader with no prior context - Cycle 3: A hostile critic looking for the weakest argument Ask each to find weaknesses, challenge assumptions, identify what is missing, and assess whether the opening earns attention. **Prompt for each cycle:** > "You are [persona]. Read the following article critically. Identify the three most significant weaknesses in the argument. Identify what is missing that a well-informed reader would expect. Assess whether the opening earns continued reading. Be direct." **On context retention:** Some models accumulate awareness across a long session and will reference their earlier critique in later cycles. Others treat each prompt as a fresh start. Persistent context tends to deepen the critique over cycles; fresh context gives genuinely independent readings. Know which you are working with. **On editorial authority:** The default is to treat the adversarial model as a critic and yourself as the editor who decides what to act on. An alternative: grant the model editorial authority to rewrite sections directly rather than just flag them. This shifts the model from L3 to L4 for the review phase. Appropriate when you trust the model's judgment on structure and clarity and want to compress the revision cycle. ### Step 6 — Revise with the critique Work through the adversarial feedback. Not every point will land, but the ones that sting are usually the ones worth acting on. If you granted editorial authority, review what changed rather than what was suggested. ### Step 7 — Final read for voice Read the finished article aloud, or ask the model to flag any sentence that does not sound like you. Your voice is not a stylistic preference. It is the signal that makes the content trustworthy. ### Step 8 — Generate the image Ask the model: "Based on this article, what would you suggest as an image generation prompt?" Let it read the full piece and propose. Review the suggestion — check that it captures the argument rather than just the topic — then tweak as needed. Hand the finalised prompt to an image generator of your choice. The image feeds into the article's metadata alongside the written content. A prompt derived from the actual argument produces something more specific and more honest than a stock photograph chosen by association. ### Step 9 — Publish Export to the appropriate format with image metadata in place. Log what the content produced and where it goes next. --- ## Maturity Requirements **Human (SFIA):** Level 3 minimum. At SFIA 3, the practitioner can apply the workflow with guidance and produce acceptable output. At SFIA 4 and above, the practitioner can adapt the workflow to complex topics and manage the adversarial review with genuine editorial judgment. The limiting factor is almost never technical. It is the quality of the original argument in step 1. A practitioner who cannot yet form a clear, hard-to-vary claim will find that AI amplifies vagueness rather than resolving it. **Model (L1-L5):** L3 for drafting and review. The AI handles structure generation, section drafting, and adversarial critique. The human drives origination, direction at each stage, and final judgment. The full spectrum: at L1 the human writes everything and AI only sense-checks; at L3 AI drafts and the human steers; at L4 the human originates the argument and grants the model editorial authority over structure and revision; at L5 the model conducts the work near-autonomously. The right level depends on the stakes, the audience, and how much of the voice needs to be irreducibly yours. --- ## Gotchas These are the non-obvious failure modes. Read before first use. **Voice erosion is silent.** If you stop originating the argument and start asking the AI to generate the claim as well as the draft, the content becomes indistinguishable from generic AI output. The one-sentence claim in step 1 is the line. As long as it comes from you, the AI cannot erode what you have not yet expressed. **Speed compresses the wrong thing.** The workflow is fast enough that it can encourage publishing before a claim is ready. Speed should compress execution time, not thinking time. Do not draft the same day you form the claim. The gap is a quality gate. **Adversarial complacency sets in gradually.** Over time, practitioners stop engaging seriously with the critique and begin approving it on autopilot. If you cannot find three genuine weaknesses in your own draft before the adversarial model does, the review step has become decoration. Rotate the reviewing model. Occasionally run the critique yourself first. **Model context retention changes the adversarial dynamic.** If your reviewing model remembers cycles 1 and 2 when it conducts cycle 3, it will build on its earlier critique. This is usually beneficial but can produce a compounding bias. If you want genuinely independent readings across all three cycles, start a fresh session for each. **Image prompts derived from topic rather than argument produce generic results.** The image generation step is only as good as the prompt. "A person using AI to write" is a topic. "A single fish navigating a fast current while others circle" is an argument. Push the model to derive the prompt from what the piece actually says, not what it is about. --- ## Business Area Impact When adopted at scale, this skill substantially reduces demand on content agencies, editorial coordinators, and junior copywriters whose primary function is turning rough expert notes into polished copy. The roles most directly affected are content coordinators and communications assistants whose work sits between expert and publication. The roles that need to upskill most urgently are subject matter experts who have never had to think about their own content production. The workflow places new responsibility on the person with the ideas. --- ## Handoff Produces a reviewed, publish-ready piece of written content plus an image prompt (and generated image where tooling permits). In an organisation with a multi-channel publishing function, output flows to a content distribution function that handles platform formatting, brand application, scheduling, and performance tracking. --- *From The Mantle Institute — mantleinstitute.com/playbook* *Human/AI Skills Matrix · Article Accelerator v1.0*