Code, camera, action 📸

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

Last month’s newsletter was a massive one — it looked like loads of you loved and found it very useful. In this week’s admittedly shorter issue:

• AI Marketing Update: AI influencer explosion

• The Stack: Using agentic AI marketing data

Onwards!

AI MARKETING UPDATE

AI influencers go mainstream 💰

Aitana Lopez posts bikini selfies from Ibiza. She talks about her skincare routine, shares gym selfies, and thanks her followers for “their love.” But she doesn’t have followers. Or pores. Or even a body.

Aitana is one of the most successful AI-generated influencers to date – a synthetic persona created by a Spanish agency, complete with a content calendar, scripted personality traits, and a rapidly growing fan base.

And she’s not alone.

From Noonoouri, a digital doll working with Dior and Valentino, to Code Miko, a fully interactive avatar streaming to thousands on Twitch and YouTube, the era of synthetic influence is already here. 

Brands are jumping in, audiences are engaging – and somewhere along the way, the lines between performance, persona, and manipulation got very, very blurry.

AI creators: A new section of digital society?

YouTube’s 2025 Culture & Trends Report landed just days ago, virtual influencers are a growing creative class.

In 2024 alone, a sample of just 300 virtual creators pulled in over 15 billion views across longform videos, livestreams, and Shorts. One billion of those views came from the U.S. alone. 

Many of these creators – like Neuro-Sama, an AI personality trained to chat with viewers and play games live – are now blurring the boundaries of entertainment, companionship, and brand promotion.

YouTube categorizes virtual creators into three buckets:

• Virtual humans (e.g. Code Miko, Miquela): AI-generated avatars with crafted personalities and narrative arcs

• VTubers (e.g. Kizuna AI, Ironmouse): Often real people using stylized anime avatars, powered by motion capture

• Character channels: Fully fictional personas 

These characters are now selling more effectively than the real-thing in some cases.

YouTube notes that many of them now participate in brand deals, product integrations, and even 24/7 livestream sales, a trend already exploding on TikTok in places like China.

Back in 2024, Aitana Lopez was claimed to be making €10,000 monthly, including campaigns with Dior, among others. She doesn’t exist, but she does endorsements, earns revenue, and posts on Instagram like any influencer. 

Miquela has similarly worked with Prada, Calvin Klein, and Samsung, and Noonoouri with several major brands, plus record deals with Warner Bros.

Creating these personas has become simpler with AI tools designed to maintain continuity for different characters you generate. We covered one method for doing this not too long ago.

To be fair, the work that goes into these projects is respectable, but it will become easier and easier. With that, AI influencers will become their own gold rush, hurtling towards saturation.

But for the time being, at least, it’s probably a sizeable nascent market.

The dark side of AI influencers

Under the surface, this is a new form of identity production, and with that comes inevitable consequences.

Not long ago, Jul Parke, a PhD candidate at the University of Toronto studying virtual influencers and racial representation, highlighted how many of the most successful avatars are built to be ethnically ambiguous, culturally trendy – and hyper-sexualized.

Aitana is a key example of this, but to be honest, you don’t need to look hard to see how ‘sex sells’ in AI influencer land. 

There’s a very real, credible argument that AI influencers represent a commodification of ethically dubious characters managed by people outside those communities.

There is an even darker spin on this, though. Investigations this year uncovered AI-generated influencers with Down syndrome, designed to attract attention under the guise of inclusion – but ultimately monetized for traffic and, in some disturbing cases, funneled toward adult content. 

Similar tactics have emerged around fake amputee and burn victim models. Exploitation dressed up as representation.

When it goes wrong, AI blows back on the brand

There are plenty of credible reasons for creating AI influencers and brand ambassadors, but it’s wise to exercise caution.

For example, in a recent study from Northeastern University, researchers looked at what happens when an influencer – real or virtual – promotes a faulty product. Who do consumers blame?

Turns out, AI influencers actually shift more blame onto the brand.

Why? Because users see them as scripted. If a virtual persona says something misleading (“This product comes with a 10-year warranty!”), it feels more manipulatory. 

Humans may get the benefit of the doubt. Machines don’t.

As the paper explains: “Organisations must recognise their accountability for the actions of AI-powered virtual influencers, as these directly affect brand trust. Selecting virtual influencers should involve not only their ability to attract followers but also a careful evaluation of potential risks to brand reputation.”

This is critical. In trying to avoid the risks of working with real people – unpredictable opinions, reputational volatility – brands may accidentally take on more responsibility than ever.

You don’t get to blame the bot when the bot ‘is’ you!

With that in mind, there are three core principles marketers need to hold onto here:

1. Declare the digital: Always disclose when you’re using AI influencers – legally, ethically, and to preserve audience trust. Transparency isn’t optional when trust isn’t optional.

2. Don’t outsource identity: If your AI avatar mimics real race, gender, or disability traits to exploit or gain – and without the lived experience or the consent of real people – you’re not innovating.

3. Augment, don’t replace: Use AI personas to extend, remix, or localize creator campaigns – not to replace them. The power of marketing lies in connection, not control. Plus, as other issues have analyzed, people can sniff out fakes that do harm through marketing.

The bottom line? AI influencers are fast, flawless, scalable, and brand-safe… on paper.

But behind the scenes, they expose uncomfortable truths about identity, authenticity, and control.

Definitely not one-way negative traffic — but with great power comes great responsibility, and all that.

THE STACK

Agentic AI is changing how marketers work with data 📊

We’re seeing a new category of tools that don’t just answer questions – they figure out what to look for next. And that shift is starting to matter in one of the most traditionally siloed parts of marketing: analytics.

Most marketing teams work with data, but few can work deeply with data. Maybe you have a dashboard. Maybe someone sends you reports.

But if you want to explore something on your own — like “Why did lead quality drop last quarter?” – you usually have to ask a data person and wait.

Agentic AI could change that. And some tools already are.

Agentic AI, in the wild

Let’s look at two examples that are already doing interesting things here: Spotter from ThoughtSpot, and Breeze from HubSpot. I have no association with either. 

Spotter, while not a marketing tool per se, is one of the most compelling early signs of what this could look like in practice. 

Let’s say you want to know whether paid LinkedIn campaigns last quarter actually drove qualified leads – but the answer lives somewhere inside a tangled mess of data tables your marketing team doesn’t normally touch. 

With Spotter, you just ask. The system parses the intent, runs the query on enterprise-grade datasets, and returns not only a clean answer but also suggestions for how to dig deeper or visualize the trend. You can go back and forth conversationally, like you would with a data analyst.

What’s smart is how it brings together technical depth and accessibility. Spotter sits on top of complex data models, but speaks in plain language. It lets marketing and growth folks work with the same datasets the analytics team sees.

Breeze, HubSpot’s new AI suite, takes a different tack. It’s fully embedded into their CRM and marketing tools. It includes:

• Breeze Copilot – a general-purpose assistant that helps you work faster across the platform. Think meeting prep, campaign recaps, and summarising contact histories.

• Breeze Agents – focused bots trained for specific workflows: content creation, social media, outreach, support. These aren’t just chat tools — they suggest actions, draft assets, and execute tasks based on your data.

• Breeze Intelligence – behind-the-scenes enrichment and signal detection. It pulls in buyer intent data, updates lead records, and shortens forms by auto-filling details.

Why data democratization is great news for marketing

What’s notable here is the shift in behaviour. These tools don’t just respond to prompts, but operate with context. 

In practice, what this means is the traditional handoffs between marketing, ops, data, and sales get blurrier. That’s mostly a good thing. 

The truth is, “marketer” isn’t one job anymore. It’s part strategist, part copywriter, part analyst, part product thinker, part community builder, part social sleuth. 

Yes, data skills are very important – and many marketers include such skills on their CV – but bridging complex data with marketing remains challenging.

Agentic AI tools will make it totally seamless as they work their way into the platforms many already use.

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.

Until next time. Happy marketing.

—The AI Marketer