Just fake it? 🎭

Hello marketers. Welcome to AI Marketing School, where we dish out the latest and greatest in AI-powered marketing.

This newsletter is in partnership with the AI Agents Summit (Sept 18-19, 2025) — a brilliant event bringing together Microsoft, Meta, Salesforce, GitHub, and more to talk about what’s actually happening with AI agents in enterprise. Check it out here.

In this week’s issue:

• AI Marketing Update: Grok for AI marketing

• The Stack: A ChatGPT Tasks experiment

• Industry Insight: AI fashion takes to the catwalk

Onwards!

AI MARKETING UPDATE

Just Grok it 🧠🚀

People are saying that "Just Google it" will soon be replaced by "Just Grok it." 

With the launch of Grok 3, Musk’s AI company xAI is positioning its chatbot as a real-time search engine – and, in his words, a “Google killer” (yes, we’ve heard that before). 

Alongside Grok 3’s improved reasoning and speed, Musk’s team unveiled Deep Search, a feature that ditches traditional indexing in favor of real-time data fetching. 

Musk is betting that Grok’s dynamic search strategy can challenge Google's long-standing dominance.

And if the hype on X is any indication, the branding is working. One user wrote:
"My Grok history is now longer than my Google Search history.”

So, how does this actually work?

How Grok 3’s Deep Search works (And why it’s different)

Traditional search engines like Google rely on a crawling, indexing, and retrieval system. 

Their web crawlers systematically scan the internet, storing vast amounts of structured data in massive, distributed databases. 

When you type in a query, Google’s algorithms match it against pre-existing indexed content, ranking results based on page authority, metadata, and user engagement signals.

Grok 3 does none of that.

Instead of maintaining a static index, Deep Search works dynamically, fetching information at the moment of a query. Musk himself describes it as "dynamic query-time retrieval.” 

Here’s what that likely means under the hood:

• Live API Calls: Instead of sifting through an old database, Grok reaches out to real-time sources – potentially through HTTP requests, X’s own API, or lightweight scraping mechanisms.

• Intent-Driven Queries: Grok parses the query for meaning (likely with an xAI-fine-tuned transformer model) and builds a focused request targeting only the most relevant, recent data.

• Real-Time Synthesis: Unlike Google, which retrieves and ranks links, Grok generates a synthesized response on the fly, discarding the data after serving it to you.

• No Cached Indexing: Google stores years of data for deep recall. Grok does not. If a page or post disappears before you ask about it, Grok likely won’t know it ever existed.

The trade-offs? Unlike Google, Grok doesn’t have years of backlogged search history to mine — only what’s available right now.

So, it’s very much oriented around current affairs. 

Can Grok actually replace Google?

Does Grok 3 really have what it takes to dethrone Google?

Having tested Grok extensively, there are some clear advantages:

• Speed & relevance: Real-time retrieval means answers may be fresher than Google’s cached search results.

• Direct synthesis: Instead of ranking a list of links, Grok provides a summarized, conversational answer.

• X Integration: If you’re searching for breaking news, trending topics, or live social discussions, Grok may already have an edge, since it directly taps into X’s firehose of posts.

But there are big challenges:

• Reliability & accuracy: Unlike Google, Grok lacks a deep search index. If live data sources are inaccurate, so is the response. It’s susceptible to manipulation, probably. 

• Transparency: Google’s search results cite sources clearly; Grok’s Deep Search does not always reveal where it pulled its information from.

• Ad model & scalability: Right now, Grok is only available to Premium subs on X. Google Search is free, ad-supported, and backed by an advertising empire. If Grok goes mainstream, how will it make money?

We can see Grok’s limitations with a simple prompt:

From this, we can see that, essentially, it’s an LLM with a real-time data feed, unlike Google or ChatGPT with search. 

Grok’s real-time model has clear strengths, but it lacks Google’s decades of archived data, robust ranking algorithms, and deep search ecosystem. For now, the two seem to complement rather than replace each other.

From an AI marketing perspective, Grok is outstanding for finding X posts that you can add to blogs and articles, finding inspiration, tracking trends, etc.

I’ve got to say, the writing is superb, albeit unpredictable enough to be a true content writing tool vs the more controllable competition. 

But that’s Grok for you!

AI agents are already changing enterprise workflows, but the big question isn’t just what they can do — it’s how businesses should be using them.

The AI Agents Summit (Sept 18-19, 2025) is a brilliant event bringing together leaders from Microsoft, Meta, Google, IBM, Adobe, and more to break down real-world use cases, integration strategies, and what actually works. If you’re thinking about AI adoption (or scaling what you’ve got), this is worth tuning into.

Get the details here or register here!

THE STACK

A ChatGPT Tasks experiment 🥼

ChatGPT Tasks is great. It allows you to schedule tasks for ChatGPT to carry out independently of you.

It’s ideal for short tasks, e.g., locating the latest news or trends. But can it handle more complex tasks, such as writing a full blog?

The answer: sort of.

Here’s the prompt I attempted with Tasks. It draws on some themes we’ve discussed in this newsletter many times, e.g,. removing overused AI terms and phrases.

Get the prompt here.

ChatGPT schedules in the task — but you’ve got to click on it and re-copy the entire prompt into the box. That’s because, by default, ChatGPT paraphrases any prompt you give it.

So, click on the task box in the chat and re-copy the prompt in, check the time/date (these can be wrong) and save.

Here’s the result. So, it works pretty well, though it evidently ignored some of the instructions. You can ask it to rewrite it.

Tasks is definitely a useful tool for accelerating content and research workflows. While you could do all of this yourself manually, it’s useful for creating a snapshot from a given time period.

Some other tested ideas:

• Use research to make social media posts based on current info, delivered to your inbox

• Find trending topics from Reddit/social media (hit or miss, though I have had good results finding very up-to-date information)

• Tracing markets, investments — this all works really well as there’s so much news available on these topics

• You can give it specific URLs or brand names — great for competitor intelligence

From the above results, packaging into content or strategies with a follow-up prompt is pretty easy.

I reckon any marketer can get some value out of Tasks. It’s no extra cost for ChatGPT premium subs and is easy to use.

INDUSTRY INSIGHT

AI fashion in the highest echelons

What happens when an AI-generated fashion campaign goes viral – and then lands its creator a deal with a luxury house?

That’s exactly what happened to Sybille de Saint Louvent, a creative director who’s been using AI to craft stunningly realistic fashion campaigns.

Initially, she was producing speculative, self-initiated AI campaigns for brands like Prada, Jil Sander, and Miu Miu – not as a stunt but as an exploration of AI’s potential in luxury marketing.

Then Gucci noticed. On Feb. 25, she announced that Gucci had officially commissioned her to create a campaign exploring the theme of duality – the same concept underpinning the brand’s Fall/Winter 2025 collection. 

The result is a dreamlike video featuring a woman staring into a mirror, only to see a man’s back reflected.

You’ve got to say it looks cool – I'm not sure what it has to do with fashion, but maybe that’s just me. 

If you can’t beat ‘em…

This strikes as ironic. Luxury fashion has always been defined by giga-budgets – million-dollar productions, A-list models, and meticulously crafted campaigns.

However, as AI-generated visuals become more refined, even high-end brands are starting to invest in the same technology that could one day democratize and disrupt their industry.

AI-powered modeling

Beyond campaigns, AI-generated models are now replacing real ones.

Fashion company Mango has already phased out some human models from its campaigns, opting instead for AI-generated avatars. Their recent teen-focused campaigns featured entirely AI-generated figures, with disclaimers noting that the visuals weren’t real.

Meanwhile, Levi’s has partnered with LaLaLand.ai to create custom digital models showcasing different skin tones, body types, and ages. 

The company claims this will offer greater inclusivity, though critics argue that hiring real diverse models would be more meaningful. 

I’m not sure ‘inclusivity’ means much when the entity you’re giving an opportunity to is an AI model…

Where does this leave AI marketing?

It’s clear that luxury brands are investing, fast-fashion retailers are streamlining, and marketing teams are adapting – all with AI. 

What’s striking is that none of the above marketing campaigns are not particularly complex to execute. Style over substance.

Not long ago, I revealed a simple tool for making AI avatars, which gets you 90% of the way there. 

If it’s good enough for Gucci then…

Hope you enjoyed this week’s issue. If you missed the last newsletter, you can read it here.

If you found it useful, please recommend it to a friend or colleague.

And don’t forget to check out the AI Agents Summit!

Until next time. Happy marketing.

—The AI Marketer