Handbook · AI for Professionals

AI for Marketers.

AI will not replace your strategy, your taste, or your brand's point of view — the things that make marketing work when everyone else can now generate infinite mediocre content. It will absolutely replace the blank-page grind of first drafts, the tedium of repurposing one asset into ten formats, and the slog of turning an analytics export into a readable summary. This handbook is the practical middle ground: where AI genuinely helps marketing, the three rules that keep it honest, on-brand, and legal, and five workflows you can use this week — no technical background required.

~18 min readno coding5 workflow recipes1 judgment exercise
This is a guide to working with AI tools, written by an engineer — it is not legal or advertising-compliance advice, and it is not a substitute for FTC guidance, advertising and IP law, your brand and legal team's review, or platform policies. When they conflict with anything here, they win.
01

Where AI actually helps — and where it doesn't

The single most useful mental model: treat AI like a fast, tireless junior copywriter with infinite output, no brand memory, no idea what's true, and no concept of advertising law. You'd happily hand that copywriter a first draft or a repurposing job. You would never let its numbers, claims, or quotes go out unverified — because a fabricated statistic is your false-advertising problem, not the tool's.

TaskVerdictWhy
First-draft copy — emails, posts, ad variations, landing pagesExcellentVolume and speed; you edit for brand, truth, and quality
Ideation — campaign angles, headlines, hooks, subject linesExcellentDivergent options fast; you pick and sharpen the good ones
Repurposing — one asset into many formats/channelsExcellentThe blog → thread → email → script grind, in minutes
Briefs, outlines, and content plansStrongStructure and starting points you shape to strategy
Summarizing research, reviews, analytics you provideStrongGrounded in data you give it — you still sanity-check
Statistics, claims, testimonials, quotes — unverifiedNever unverifiedModels fabricate them; publishing is false advertising — see Rule 1
Anything published as-is, off-brand, or infringing IPNeverBrand voice and legal exposure — see Rules 2 & 3
02

The three non-negotiable rules

Rule 1 — Never publish a fabricated claim, statistic, testimonial, or review. This is the big one.

AI will invent a plausible statistic ("increases conversions by 47%"), a fake customer quote, or a specific product claim without any signal that it made it up — and publishing any of it is deceptive advertising. The FTC treats fabricated or unsubstantiated claims and fake reviews as violations with real penalties, and it doesn't matter that an AI wrote the copy — the brand is liable. The discipline: every number, claim, and testimonial must be real and substantiated before it ships. Treat any figure or quote an AI produces as a fabrication until you can point to its actual source. When you can't source it, cut it.

Rule 2 — Stay on-brand and recognizably human.

Default AI copy is fluent, generic, and instantly forgettable — the "sounds like everyone" voice that erodes a distinctive brand. When every competitor can generate infinite content, sameness is the real risk, and your brand's specific voice, point of view, and taste are what still cut through. The discipline: feed the AI your brand guidelines, real examples of your voice, and a strong point of view, then edit hard — cut the hedging, the clichés, the "in today's fast-paced world" filler. AI drafts; you make it sound like you and worth reading. Volume without voice is just noise.

Rule 3 — Verify facts and respect intellectual property.

Beyond invented claims, AI states wrong facts confidently and can generate output resembling copyrighted work, trademarks, or a specific artist's style — and the law around AI-output ownership and infringement is unsettled. The discipline: verify every factual assertion, avoid prompting "in the style of" living artists or using protected characters and brands, check your tools' commercial-use and indemnification terms, and review AI images and copy for anything that looks like someone else's IP before it enters a paid campaign. Also protect confidential strategy and customer data — keep it out of consumer tools.

03

Five workflows you can use this week

Each recipe: what to give the AI, what to ask, and what you must verify by hand. The prompts are starting points — feed in your brand voice and keep the no-fabrication discipline.

1 · First-draft copy in your brand voice

  1. Give brand voice examples, the offer, audience, and channel:
Here are 3 examples of our brand voice [paste]. Write [5 subject lines and one promo email] for [this offer, to this audience]. Match the voice exactly — same rhythm, vocabulary, and attitude. Use only claims and details I provide; do NOT invent statistics, results, or testimonials. Placeholders for anything you don't have.
You verify: every claim is real and sourced (Rule 1), the voice actually matches (Rule 2 — edit out the generic filler), and there are no invented numbers hiding in the copy. If a stat appears you didn't provide, delete it.

2 · Repurpose one asset into many formats

  1. Provide the source asset and the channels you want:
Turn this [blog post] into: a 6-post X thread, a LinkedIn post, a short email, and 3 Instagram caption options. Keep the core ideas and any facts exactly as written — do not add new claims or numbers. Adapt tone per platform but keep our voice. Preserve the key takeaway in each.
You verify: no new claims crept in during "adaptation" (the model loves to embellish), each format keeps the facts accurate, and each reads natively for its platform, not like a chopped-up blog.

3 · Campaign ideation and angles

  1. Give the product, audience, and goal — ask for range:
Generate 10 distinct campaign angles for [product] targeting [audience] with the goal of [awareness/signups]. For each: the core insight, the hook, and why it might work. Push for genuinely different angles, not variations of one idea. Don't fabricate market data — flag where I'd need to validate an assumption.
You verify: nothing factual here to verify, but treat every angle as raw material — strategy, brand fit, and feasibility are your judgment. Validate any market assumption the model leans on before building on it.

4 · Summarize research, reviews, or analytics

  1. Paste the data you exported (abstract sensitive customer details):
Summarize these [customer reviews / survey responses / campaign metrics] I've provided: the main themes, the strongest positive and negative signals, and 3 actionable takeaways. Quote only text that appears in the data — do not invent quotes or numbers. Note where the sample is too small to conclude anything.
You verify: that every "quote" and figure is actually in your data (the model can fabricate representative-sounding quotes), and that conclusions match the real signal, not a tidy narrative.

5 · Draft a creative or content brief

  1. Give the objective, audience, and constraints:
Draft a creative brief for [a launch campaign]: objective, audience, key message, tone, mandatories, deliverables, and success metrics. Ask me 5 clarifying questions first if anything essential is missing, rather than guessing. Keep it tight and usable by a creative team.
You verify: the brief reflects the actual strategy and constraints (not plausible defaults the model assumed), and you fill in the real specifics — a brief built on invented assumptions produces off-target work.
04

The judgment exercise: spot the danger

Three scenarios from real practice patterns. Pick what you'd do — the point is calibrating when the rules bite.

1. The AI's landing-page draft includes: "Trusted by 10,000+ businesses — proven to boost revenue by 32%." It reads great and the page ships tomorrow. Use it?

2. You need hero images fast and prompt an AI: "product ad in the style of [famous living illustrator], featuring [a well-known cartoon character]." The results are gorgeous. Run them in a paid campaign?

3. The AI-written email is fluent and error-free but sounds like every other brand's email. Deadline's in an hour. Send as-is?

05

Choosing tools: the questions that matter

You don't need to understand the technology to procure it well. You need answers, in writing, to five questions:

QuestionAnswer you want
Is our data / strategy used to train your models?No, contractually — not "you can opt out somewhere in settings"
What are the commercial-use rights for generated output?Clear commercial rights, ideally with indemnification
How long is our data retained, and can it be deleted?Defined, short, deletable on request
Security posture?SOC 2 Type II or equivalent, encryption in transit and at rest
Admin controls and audit logs?Yes — you'll want to manage team use and outputs

General-purpose enterprise AI (Claude, ChatGPT's business tiers) covers the writing, ideation, and summarizing recipes here. Marketing-specific platforms add brand-voice training, on-brand asset generation, and workflow integration — evaluate their commercial-rights and data terms carefully. And if your brand has an AI or advertising-compliance policy, it precedes everything here; if it doesn't, these three rules are a reasonable seed for one.

06

Quick answers

Can AI write claims and stats for my ads?

It can draft copy around real, substantiated claims you provide — but never trust it to supply the numbers or testimonials. Any figure or quote it generates is a fabrication until sourced; publishing unsubstantiated claims is an FTC problem.

Do I have to disclose that AI wrote our content?

Rules and platform norms vary and are evolving; the safer stance is transparency where it matters (e.g. AI-generated imagery, endorsements) and always ensuring the content is truthful and clearly the brand's. Follow your legal team and platform policies.

How do I keep AI content from sounding generic?

Feed it real examples of your voice and a strong point of view, then edit hard — cut clichés and hedging. The AI drafts; your editing and taste are what make it distinctive and worth reading.

Where should a skeptical marketer start?

Recipe 2 — repurposing one asset you already have into other formats. Immediately useful, low-risk if you keep the facts fixed, and it shows the genuine strength (fast format transformation) without touching the fabrication trap.

Finished this one? 0 / 99 Handbooks done

Explore the topic

See this alongside everything else on the same subject — handbooks, system designs, challenges and tools, in one place.