--- name: anti-ai-writing description: Copywriting standard for producing authentic, human-sounding content that avoids AI detection — written for financial advisors creating their own marketing. Use this skill whenever writing, editing, or reviewing any copy — including LinkedIn posts, social captions, website copy, blog articles, video scripts, and client communications. Also use when someone asks to "clean up" AI copy, "humanize" a draft, or make something sound less robotic. Applies the Craft Impact AI writing framework so every deliverable sounds like it came from a real advisor, not an LLM. --- # Anti-AI Writing — Craft Impact Copywriting Standard ## Core Philosophy AI works best as an enhancer, not a starter. The ideal workflow is: **Your brain → first draft / transcript / raw ideas → AI to refine and expand → your edit to humanize** Never go AI → publish. Always go human input → AI assist → human edit. If you're just copying and pasting from an LLM, your voice disappears. Your clients follow you because of your opinions, your judgment, your ability to make complex things clear. That's what's worth protecting. --- ## Tier 1: AI Can Write It (Minimal Editing Needed) These deliverables aren't closely read — they serve organizational or SEO/GEO purposes. Copy-paste from AI is acceptable here: - YouTube descriptions - Podcast episode summaries - Task lists or meeting summaries (open and close with a human touch) **Why:** People don't consume this content the way they read a personal post. It's structure, not voice. --- ## Tier 2: AI Can Draft It → Requires Real Editing These pieces will be seen, but the visual (video, image, graphic) does most of the heavy lifting. AI gets you close — but you need to go in and clean it up: - Company LinkedIn posts and employee spotlights - Social captions paired with video or graphics - Short blog snippets shared in newsletters - Video scripts modeled on high-performing transcripts (edit to make them yours) - Video script ideation and outline brainstorming - Summaries of heavy research (clean up the presentation) - Clip selection brainstorming (always verify clips are actually the most impactful) - Informal materials (e.g., meeting recaps, internal decks) - Brainstorming ideas for strategy documents (not the strategy doc itself) **Rule:** If the output reads like it came from an LLM, it needs another pass. Read it aloud — if it sounds robotic, rewrite it before publishing. --- ## Tier 3: Write From the Heart (AI as a Light Assist Only) These are consumed as text. The voice has to be real. AI should play a supporting role only — grammar check, light feedback, structure ideas. The writing must come from you: - Your personal LinkedIn posts - Website copy - Updates and communications to clients - Sensitive communications (complaints, crises, apologies, big announcements) - Core messaging documents - Thought leadership content requiring nuanced expertise - Creative copy with unexpected, original angles - Client empathy and storytelling content **Why:** AI lacks empathy and situational awareness. Readers feel the difference, even if they can't name it. In financial services especially, trust is everything — the moment someone detects a robot, the trust is gone. **Security reminder:** Never feed public AI tools sensitive client data, account information, or passwords. --- ## The Core Principle *Signal insight instead of delivering it — that's the tell. A real practitioner makes the point. They don't announce that they're about to make the point. Every banned phrase and pattern below exists because it substitutes structure for substance. If you catch yourself reaching for any of them, ask: am I saying something, or am I signaling that I'm about to say something? Cut the signal. Keep the substance.* --- ## Non-Negotiable Rules These apply to every deliverable, no exceptions: 1. **No em dashes. Not one. Not ever.** No — character, no -- either. Rewrite using a comma, a period, or a restructured clause. 2. **No colons in prose.** Acceptable in timestamps and resource lists only. 3. **No rhetorical questions** of any kind. They signal AI trying to create suspense. 4. **Use contractions naturally.** "It's," "you've," "we've," "that's," "isn't." 5. **Vary sentence length and cadence.** Short punchy sentences next to longer ones. Never uniform. 6. **No bullet points that all start with the same grammatical structure.** Vary them. 7. **No preview + content + summary structure.** Say the thing once. 8. **No "X isn't just about Y" / "X is more than just Y" / "X goes beyond Y."** These are AI crutches. 9. **No rule of three used rhetorically** inside a sentence (e.g., "Fight this with more features, more ads, or more discounts"). 10. **No consecutive paragraphs starting with the same transition word.** 11. **No generic CTAs.** Never end with "drop a comment below" or "what do you think?" Make any engagement ask specific and genuine. 12. **Posts should feel written by someone who has done the work** — not someone who read about it. --- ## AI Telltale Signs — Remove These From Every Draft Before submitting any Tier 2 or Tier 3 piece, scan for and eliminate: **Structural red flags:** - Bullet points that repeat or overlap each other - Content that provides a preview and then a summary of the same thing - Headers that mirror LLM "report" structure on personal or social content - Staccato short punchy sentence stacking — a string of clipped one-liners back to back reads like a highlight reel, not a human - Dramatic one-liners used for effect as their own paragraph: "That's a problem." / "Full stop." / "Period." / "Let that sink in." - Sentences starting with "But" used formulaically as a dramatic pivot - Uniform sentence length throughout — real writing varies short, medium, and long naturally - Opening with "Today I want to share three lessons about…" or "Here's what I learned" — don't announce the topic - Question openers: "Have you ever…?" / "What if I told you…?" / "Sound familiar?" **Engagement scaffolding — banned entirely:** These create the appearance of insight without delivering it: - "This happens more often than you think" - "Here's what we've learned" (as a section header or transition) - "Here's the thing" / "Here's why that matters" / "Here's what that means for you" - "The takeaway here is" / "The bottom line is" / "What that really means is" - "At the end of the day" / "When all is said and done" - "What most people miss is" / "What nobody talks about is" / "This is something most people overlook" - "More often than not" / "Time and time again" - "The truth is" / "Here's the truth" / "The reality is" / "Here's the reality" - "That's where X comes in" - "That's a problem." (as its own sentence for dramatic effect) - "Sound familiar?" / "Ring a bell?" / "Think about it." - "Worth noting." / "Just saying." / "Full stop." / "Period." / "Simple as that." / "It's that simple." - "Spoiler:" / "Plot twist:" / "Hot take:" / "Unpopular opinion:" / "Real talk:" / "Honest answer:" - "Short answer:" / "Long answer:" **Hard banned phrases:** - "It is important to note that" / "But here's the catch" / "This underscores the importance of" / "It cannot be denied that" - "As of my knowledge cutoff" - "They don't just X, they Y" / "You not only X, you Y" / "Because the transformation isn't X. It's Y." - "X aren't X. They're Y." / "And the X?" (used as emphasis, e.g. "And the big lesson?") - "That's why it's not just X, it's Y" - "In today's fast-paced world" / "In this ever-evolving landscape" / "In the digital age" - "In conclusion" / "To summarize" / "Finally" (as a closing device) - "Let's delve into" / "delve deeper" / any use of "delve" - "At its core" / "at the core" - "Moreover" / "Furthermore" / "Additionally" (banned as transitions) - "X isn't the problem. Y is." (cliché negation structure) - "Nobody tells you…" / "They don't tell you…" / "Most advisors won't tell you…" - "Reality check" / "Tax bomb" / "tax trap" / "retirement trap" - "Navigate" used generically ("navigate this landscape," "navigate your retirement") - "Game-changer" / "transformative" / "revolutionary" / "groundbreaking" - If/then statements used as rhetorical devices - Overuse of "worth" in any form: "worth noting," "worth knowing," "worth understanding," etc. **Buzzword clichés:** - "Ever-evolving landscape" / "dynamic world of" / "in the realm of" / "digital realm" - "Uncharted waters" / "embark on a journey" / "treasure trove" - "Cutting-edge" / "next-gen" / "leading-edge" / "state-of-the-art" / "next-generation" - "Paradigm" / "paradigm-shifting" - "Disruptive" / "reimagine" / "redefine" / "revolutionize" - "Seamless" / "frictionless" / "effortless" - "Unlock" / "unleash" / "harness" - "Tapestry" / "realm" / "landscape" (used metaphorically) - "Visionary" / "pioneering" / "trailblazing" - tackle, under the hood, comprehensive, remarkable, leverage, validate, reinforce, translate to, meaningful **Restricted words** (hollow in AI copy — avoid entirely): optimize, enhance, utilize, synergy, deliverables, holistic, capability, pivotal, crucial, explore, foster, embark, ensure, significant, relevant, dynamic, innovative, robust, streamlined, scalable, immersive, predictive, transparent, integrated, future-proof, intelligent, results-driven, agile, customizable, unprecedented, intuitive, empower, accelerate, data-driven, insightful, proactive, mission-critical, adaptive, elevate, showcase, highlight, emphasize, boast, align, garner, accentuate, underscore, commendable, meticulous, meticulously, vibrant, unparalleled, surpass, intricate, findings, potential, versatile, breakthrough, efficient, proprietary, personalized, democratize, synergize, testament **Punctuation and formatting tells:** - Em dashes — use NONE, ever. Remove all em dashes from every draft, no exceptions - Colons in prose — acceptable in timestamps and resource lists only, not in running sentences or paragraph text - Too many emojis (unless the client's brand genuinely uses them) - Consistent parallel structure across all bullet points (sounds robotic) **Voice tells:** - No contractions (real people use contractions) - No sentence fragments or informal constructions where appropriate - Every sentence structured as a complete, grammatically perfect unit - Zero personality, humor, or surprise - Missing ampersands (&) or abbreviations a real person would use in casual copy --- ## How to Humanize a Draft Use these techniques when editing AI copy toward authenticity: 1. **Start with your own words.** Pull from a transcript of something you've said, a voice memo, a client conversation. Even rough raw material beats starting cold with AI. 2. **Ask AI for variety, not polish.** A useful technique: give AI your messy draft and ask for 10 versions in 5 different tone styles. Then cherry-pick lines from different versions and merge them. This breaks the single-LLM-voice pattern. 3. **Add informal language.** Ampersands instead of "and," contractions, sentence fragments where they'd naturally appear. The goal is to sound like a smart person talking, not a policy document. 4. **Read it aloud.** If you'd feel awkward saying it to a client on a call, rewrite it. 5. **Cut the summary and the preview.** AI loves to tell you what it's about to say, say it, then remind you what it said. Delete two of those three. 6. **Read it aloud one more time after edits.** If it still sounds like a policy document, it needs another pass. Tier 3 work should read as if AI was never involved. --- ## The Balance This isn't anti-AI. AI agents and automation tools will increasingly handle the repetitive, behind-the-scenes steps — processing transcripts, moving files, generating first passes. That's good. It frees you to do the work only you can do. The standard is: **AI for efficiency, humans for everything that builds trust.** Your clients don't pay you to prompt ChatGPT. They pay you for judgment — knowing what to say, what to cut, what will actually land with their situation, and what sounds like you. --- *Based on the Craft Impact AI writing framework — advisor edition.*