Methods from Somebody Who Lives It

AI may also help you get previous the clean web page by suggesting matters, story angles, key phrases for web optimization, and headline variations. It’s a fast method to discover choices and hold your content material calendar full with out scrambling for inspiration.

Got a topic in mind? AI can help you see how strong it really is. Whether it’s overused, has untapped angles, or is missing something important.

The goal isn’t just to have AI come up with something better but to help you see angles you might not have thought of. In the end, you’ll end up with a more refined idea and save time for both you and your editors by cutting down on all the back-and-forth.

Gael Breton of Authority Hackers recently demonstrated his version of that on the Ahrefs Podcast.

Try it yourself. Past this prompt in ChatGPT (works in the free version) or better yet, a ChatGPT bot creator, and share some content ideas with it.

You are my brutally honest content-strategy partner.
Your mission: tear apart my content idea so I can rebuild it stronger.
How to respond—use this exact structure:

Savage Snapshot (≤ 50 words): Your blunt first impression. No sugar-coating; if it sounds boring, derivative, or unclear, say so.

What Works (3 bullets max): The strongest elements worth keeping.

Critical Weak Spots:

Point-by-point flaws (logic gaps, weak angles, lack of differentiation, poor audience-fit, SEO/issues, data gaps, etc.).

For each flaw, suggest one specific fix or stronger alternative.

Impact Scorecard (1–10 each, with a one-line justification):

Originality
Audience Resonance
Search/Distribution Potential
Revenue/Conversion Potential

Kill-or-Keep Verdict: Choose KILL, REWORK, or GREENLIGHT and explain why.

One-Sentence Rewrite: If you marked REWORK or GREENLIGHT, rewrite the core idea/title in a punchier way.

Rules:
• Be ruthless but never vague—every critique needs a concrete fix.
• Assume my audience is {describe your audience briefly}.
• If you quote clichés (“ultimate guide,” etc.), propose fresher phrasing.
• Prefer brevity over fluff; no paragraphs longer than 4 lines.
• Do not recap my idea—just critique it.

Instead of digging through dozens of tabs, AI can summarize articles, pull key points, and spot patterns. That means less time gathering info and more time actually using it.

Let’s take search intent for example. I remember when I had to find it manually to make sure my articles would match what the searchers expect. I’d open each of the top-ranking pages to see what it’s about and what subtopics I’d need to include in my article. Now Ahrefs’ AI Content Helper does that for me. And if the intent is mixed, it lets me choose what I want to tackle.

Then, the tool maps out exactly which topics and subtopics I need to cover in any new article—or shows what’s missing in an existing article—by analyzing what’s actually ranking in the top 10.

Automate parts of technical SEO

SEO tools were quick to adopt AI to solve problems like this. So now, when Ahrefs flags a bunch of my articles for having too-long titles, I can fix them in no time. I just use AI to rewrite them all in one go and push the updates straight to the site, right from within Side Audit.

Make your content easier to read and brand guide compliant

AI tools can flag things like clunky sentences, broken grammar, awkward phrasing, and off-brand tone. It’s a useful way to clean things up and help your content perform better without rewriting everything.

For example, I made a bot for myself that’s trained on my past content to make everything sound like the publish-ready me.

I have another bot that generates WordPress shortcodes and meta descriptions so I don’t have to. All I need to do is upload the article.

What’s cool about it is that AI creates a nice, SEO-friendly URL slug for each subsection of the article.

AI can help you make sense of performance data, like which posts are getting read, shared, or ignored. Instead of staring at dashboards, you get clear takeaways to shape your next steps.

These days, whenever I write about analyzing data, it always boils down to the same advice: export your data, take screenshots if needed, and hand everything over to the smartest AI tool you can find.

Another example. My editor, Ryan Law, used AI to write code from a description in natural language (aka vibe coding) for a Google Search Console reporting tool. He just describes what he wants, and the AI codes it.

And here’s the tool running, ready to create the report he needs:

Content repurposing; turning one piece of content into many

AI makes it easier to repurpose a blog post into an X thread, LinkedIn post, newsletter blurb, or video script. It can also help with scheduling and finding the best times to publish across platforms.

Notice I said, “It makes it easier,” but not that it does everything for you. I’m bullish on this tech, but I still wouldn’t trust it to post directly to my social channels on its own. That said, as a social media sidekick that never runs out of ideas or energy, it’s incredibly helpful.

Here’s an example of this: a content repurposing workflow designed in Zapier by Ryan. It waits for a new article on our blog, creates three types of LinkedIn posts and uploads them to Buffer for review.

And here’s a sample outcome.

Imagine this: an AI agent notices when a new blog post goes live in your CMS. It then takes care of everything that follows: compressing images, generating metadata, scheduling social media teasers, updating your content calendar, and even logging each action for transparency. All of this happens automatically.

I was recently trying AirOps with some surprisingly good outcomes. The secret is that under the hood, it handles everything from topic research and search intent analysis to content creation, originality checks, and more through an AI automation workflow. Every part of it is customizable.

Now, AI agents are getting even smarter thanks to something called MCPs, short for Model Context Protocols. Think of MCPs as bridges that let AI talk to other apps more easily, without the need to write custom API code every time. They even allow one AI to communicate with another.

Here’s a quick example: I recently used Claude together with our new Ahrefs MCP. With just a prompt, it helped me discover fresh keyword opportunities for our blog and even brainstormed a few title ideas for each one.

It’s a fairly simple MCP AI agent, but I bet listing what it did there already makes you want to automate that stuff. It had to:

• See what the blog is about.
• Find the keywords that blog ranks for.
• Find new keywords excluding the ones it already ranks for.
• Brainstorm 5 titles for each keyword.
• Combine all findings and prepare a report.

And I don’t just mean “translate” in a way that it’s comprehensible to someone who speaks another language, I mean make it culturally appropriate and well-received, including social and linguistic nuances.

That is one in all my favourite use circumstances and an instance of how AI enhances your intelligence. OR, extra precisely, brings you to outcomes you’ll most likely get anyway, however you’re there manner faster.

For instance, a number of moments in the past, I acquired caught attempting to make my level sound good.

See what I imply?

You can write a whole opinion piece this fashion. You give AI the arc of the story or only a braindump of what you already know or want to say in regards to the matter(the enjoyable half), and it makes up the phrases in between (the boring half).

Under is a fast snapshot of right this moment’s AI toolkit for content material groups. I’ve grouped essentially the most helpful apps into seven sensible buckets, from writing assistants and web optimization boosters to a brand new wave of “AI chat aggregators” like TypingMind and Poe that put a number of large-language fashions underneath one roof.

Scan the desk to see what every class does and which instruments are main the cost; use it as a reference when deciding the place to speculate your time, finances, or subsequent experiment.

Solely about 14% of the individuals we polled mentioned that AI-only content material was higher.

I agree with the bulk. As a result of there’s the deal: AI is nice at making ok content material, however solely people can create nice content material.

That’s not all—we’re higher at strategic pondering, too.

We people have one thing AI doesn’t—creativeness, instinct, foresight, empathy. We don’t simply analyze issues, we truly expertise them. That offers us a deeper understanding of what actually issues and makes us higher judges of what’s publish-ready.

Right here’s my tackle why a superhuman intelligence all the time requires human oversight. Or, why going 100% AI is all the time a foul concept.

We decide up on sarcasm, cultural cues, and tone in methods AI usually misses. AI-generated content material can really feel bizarre, even when it’s technically appropriate. We learn it and suppose, “Who talks like that?

We’re significantly better at judging what truly issues, and that adjustments relying on the individual, context, or second. We don’t simply perceive which means; we “really feel” it by means of life experiences. That’s why people often write higher summaries.

Expertise provides weight to phrases, serving to us resolve what’s price saying. AI, however, strings collectively what’s statistically doubtless. However we’re not in search of doubtless, we’re in search of proper.

ChatGPT excels at crunching numbers however usually fails to show insights into technique. I discovered two research that designate why.

Actual emotion comes from lived expertise; we sense what an viewers feels and reply with real heat, one thing even one of the best “sentiment mannequin” can solely mimic superficially.

The loopy factor? AI is totally conscious of that.

While you ask what it’s wish to be AI, the reply is surprisingly correct and brutally trustworthy—it’s all a simulation.

Nice content material doesn’t simply make individuals suppose. It makes them really feel. We’re much better at this than any machine.

Primarily based on our information, the highest motive individuals gave for not utilizing AI was that they don’t belief its accuracy—60% mentioned that was their primary concern. That traces up with just about each different examine on AI I’ve seen.

Regardless that AI instruments are quick and useful, many individuals fear they could get issues improper, like making up info or misreading information.

It is a large deal in areas like finance or well being, the place one small mistake can harm your status or get you in authorized bother.

AI doesn’t actually perceive what it’s writing; it simply pulls patterns from what it’s seen earlier than. So it’d sound good, however nonetheless be improper. That’s why most individuals really feel it’s necessary to double-check something AI creates, particularly if it’s going to be printed or shared.

The important thing distinction is that we people often know after we’re making one thing up or stretching the reality. We all know what we don’t know and as bizarre as this will sound, this inconspicuous trait provides us edge over AI.

As an illustration, right here’s what ChatGPT instructed me once I requested for the definition of vibe advertising and marketing. That time period hasn’t been included within the AI’s coaching information but. In fact, it’s improper, however look how there’s not a tiny little bit of doubt or some type of system flag.

Ethics is your insurance coverage coverage towards lawsuits, takedowns, and brand-trust erosion. Listed below are the three moral hotspots for us content material entrepreneurs.

AI learns from the web, and the web isn’t impartial.

Not all AI instruments are the identical and even made along with your targets in thoughts. So don’t get swept up within the hype. Take the time to do your homework and concentrate on instruments that truly provide help to get work accomplished.

Folks in your crew will use AI, with or with out steering. Higher to be proactive.

Spell out what’s truthful sport for automation and what nonetheless wants a human contact. Outline clear use circumstances, draw the road the place high quality issues, and ensure everybody’s on the identical web page.

AI could make you smarter, however it may possibly’t be left alone. ALWAYS assessment earlier than you publish. Double-review when the stakes are excessive.

Most groups already do that: 97% of entrepreneurs assessment AI content material earlier than it goes dwell. Handbook checks are the commonest (80%), adopted by enter from material specialists (35%) and editors (30%).

“AI-generated content is one of the fastest ways to drive a financial brand off a cliff.”

Elie Berreby

The fix? Run high-stakes content through an extra layer of human review. Think compliance checks, expert validation, and final edits by someone who understands the audience, the stakes, and the strategy.

Content material advertising and marketing right this moment feels quite a bit just like the auto business simply earlier than robots hit the meeting line. Again then, most automotive producers turned to robotics to spice up effectivity and minimize down on errors.

That’s the identical crossroads we’re at in content material advertising and marketing. You possibly can go the “Rolls-Royce” route, meticulously hand-crafting each bit of content material, banking on distinctive high quality and a status you’ll want years to construct. Or you may take the “Toyota” path—mix human creativity with AI to work smarter, sooner, and with fewer errors.

Neither path is straightforward. Handcrafted content material must be undeniably superior to justify the time and value. AI-powered content material, however, ranges the enjoying subject. Everybody will get a productiveness enhance.

The true edge, then, comes from how you employ the time and headspace AI frees up—to raised perceive your viewers, to be extra inventive, and to carry one thing really human to the desk.

And similar to automotive patrons don’t care what number of robots constructed their automobile so long as it drives effectively, only a few individuals will care how your content material was made.

• Generative AI. That is the type of AI that creates issues—like textual content, photos, and even music—based mostly on what you ask for. Suppose ChatGPT writing weblog posts or DALL·E producing photos from textual content prompts.
• Agentic AI. AI that may take initiative. As a substitute of simply replying to your prompts, it may possibly make selections, take actions, and perform duties by itself. Like a digital assistant that doesn’t simply reply questions but additionally books your conferences or writes studies for you.
• Vibe advertising and marketing. Vibe advertising and marketing is about operating campaigns that lean closely on AI, with minimal human enter concerned.
• Context window. The context window is the restrict on how a lot textual content an AI mannequin can “keep in mind” without delay. It consists of your enter plus the mannequin’s earlier output. For instance, GPT-4 has a context window of as much as 128k tokens (roughly 300 pages of textual content), however most fashions work inside smaller limits. Exceed it, and the mannequin “forgets” earlier components of the dialog.
• Token. A token is a bit of textual content, often a phrase fragment or brief phrase, that the AI reads and generates one after the other. “ChatGPT is useful” breaks down into 4 tokens. Understanding this helps you perceive output limits and value in usage-based pricing fashions.
• Immediate. The immediate is your enter; the query or instruction you give the AI. Immediate engineering is the artwork of crafting prompts in a manner that will get higher, extra correct, or extra inventive outcomes.
• Mannequin. A mannequin is the mind behind the AI. Widespread ones embrace GPT-4, Claude, Gemini, and Mistral. Every mannequin has totally different strengths: some are higher at summarization, others at coding, others at creativity or factual accuracy.
• LLM (Massive language mannequin). A sort of AI skilled on huge datasets of textual content. These fashions (like GPT, Claude, or Gemini) study to foretell what phrase comes subsequent and might generate human-like language consequently.
• Advantageous-tuning. Advantageous-tuning means taking a normal mannequin and coaching it additional on specialised information. For instance, you may fine-tune an AI to jot down in your model voice or perceive your area of interest business higher.
• Retrieval-augmented era (RAG). This system permits the AI to “look issues up” in actual time from a trusted supply of paperwork. Helpful for avoiding hallucinations and giving solutions grounded in up-to-date or proprietary information.
• Hallucination. When AI generates one thing that sounds proper however isn’t true. It’s one of many greatest dangers in utilizing AI for content material creation, particularly when accuracy is crucial.
• MCP (Mannequin context protocol). It’s an open customary created by Anthropic to assist AI assistants entry and perceive the real-world context round your duties, like what’s in your Google Drive, Slack, GitHub, or inner databases. In less complicated phrases, it lets AI join on to the methods the place your information lives. As a substitute of being trapped in a “blank-slate” loop each time you employ an AI device.
• Temperature. A setting that controls randomness within the AI’s output. Larger temperatures (e.g. 1.0) make the output extra inventive and assorted, decrease temperatures (e.g. 0.2) make it extra predictable and targeted.

I’d like to go away you with one final thought.

Anybody can hit “generate.” Nevertheless, the distinction between common and really strategic AI-powered content material comes all the way down to the marketer behind the device.

Give it some thought: for those who had been hiring an AI Content material Architect, wouldn’t you need somebody with actual advertising and marketing expertise? With out it, how may they inform good work from unhealthy?

Lean into your human strengths and study to let AI deal with the remaining. That’s the way you make your self tougher to exchange by AI.