AI can automate creator management workflows. But first, it has to actually watch the videos.

By Megan Duong, CEO and Co-Founder of Plot

AI is moving quickly into creator marketing. Everyone wants technology that can find creators, vet them, track campaigns and tell you who to work with next.

And the investment is following. IAB projects U.S. creator ad spend will reach $44 billion in 2026. Nearly half of creator ad buyers already consider creators a “must buy,” yet identifying the right creators remains brands’ biggest challenge. Three in four brands are also using or planning to use AI for creator marketing-related tasks.

I understand why. Creator marketing has become a core channel for brands, but so much of the work behind it is still incredibly manual. Teams spend hours searching for creators, opening profiles, watching video after video, checking old partnerships, trying to understand whether someone genuinely fits the brand, and then tracking what actually gets posted.

AI should be able to take a lot of that work off marketers’ plates.

But there’s a problem I don’t think we talk about enough: the quality of an AI workflow is constrained by the information it can see.

And in creator marketing, most of the valuable information lives inside the content itself.

The most valuable signal might never be tagged

We saw this firsthand with Benefit Cosmetics.

One creator with 6.2 million followers had a Benefit product sitting on her dresser across her videos. She never tagged Benefit once. Over three months, she featured the brand seven times, and those videos generated more than 32 million views.

Benefit didn’t know she existed.

And she wasn’t an anomaly. We found that roughly 40% of the creators talking about Benefit weren’t tagging the brand.

Think about what that means. A brand can have people organically using and advocating for its products at enormous scale and still be completely blind to it if its technology is only looking for explicit mentions.

A caption might not mention your brand. A creator might never tag you. But your product could be sitting on her bathroom counter in five videos. She could be recommending it out loud, wearing it in a GRWM or repeatedly featuring it as part of her routine.

A marketer watching those videos would see those signals immediately.

AI needs to be able to see them too.

The creators worth finding aren’t always the ones your database would surface

We saw another version of this when we analyzed more than 6,000 creator posts around Gap’s recent collaborations.

One of the standout posts came from an account with just 427 followers. The video generated 1.1 million views and more than 120,000 likes.

She didn’t tag Gap.

If you were sourcing creators based primarily on follower count or tracking brand mentions based on tags, you could miss her twice.

This is why I think the conversation around AI in creator marketing needs to start one step earlier than automation.

The industry is focused on making AI smarter. But intelligence isn’t only about the model. It’s also about what the model can see.

Give AI a traditional creator database and it can automate searching a traditional creator database. It can filter by follower count, engagement rate, location or category faster than a person can.

But those are still the same signals marketers have had for years.

If AI can actually understand millions of videos — the visuals, spoken words, on screen text, captions and the context around them — it can start reasoning over the information marketers have historically had to uncover manually.

Intelligence first. Automation second.

Before asking what AI can automate, marketers should ask what information it is capable of understanding.

• Can it see what products appear inside a video even when they aren’t tagged?
• Can it understand what a creator actually talks about, not just the category assigned to their profile?
• Can it recognize that someone has organically used the same brand repeatedly over time?
• Can it understand the communities, interests and behaviors that show up across a creator’s content?
• Once AI can understand that context, the workflows themselves start to change.

Sourcing gets more precise. Instead of starting with filters like “beauty creator, New York, 50,000+ followers,” a marketer can search for something much closer to an actual brief: Find NYC creators who recently started running, organically talk about skincare and are training for their first marathon.

Vetting gets deeper. AI can research what someone has actually said, shown and promoted, look across previous partnerships and organic content, and surface potential conflicts or signals of genuine affinity.

Tracking gets more complete. Instead of measuring only the posts where a creator remembered to tag the brand, teams can understand paid, gifting, seeding and organic advocacy — including the content where the product appears but the brand is never mentioned.

That’s a very different version of automation than simply making database searches faster.

Automate the research, not the judgment

There’s still an important line to draw.

Finding a signal and deciding what to do with it are two different things.

Culture is nuanced. Brand fit is subjective. A creator can check every box on paper and still feel completely wrong for a brand. Marketers should still decide who they want representing them, what ideas are worth pursuing and when something simply feels right.

I spent my career in marketing before building Plot, and I don’t know many marketers who got into this industry because they wanted to spend their day searching databases and watching hundreds of videos to populate a spreadsheet.

They want to understand people. Find the insight. Build the creative idea. Know when something feels right for the brand.

That’s why I’m incredibly bullish on AI helping marketers work faster and analyze more information. I’m much less interested in using it to replace the craft of marketing.

As AI gets embedded into more creator workflows, there will be enormous pressure to automate simply because we can.

Before doing that, marketing leaders should ask one question:

What information is the AI actually seeing?

If it’s working from profiles, captions and tags, we may just be automating yesterday’s version of creator marketing.

But if AI can actually watch the content — what creators say, show, use and do — it can take on the tedious work of finding, researching, vetting and tracking creators while giving marketers something much more valuable: a fuller picture on which to make the call.

The goal isn’t to automate taste or judgment.

It’s to automate the work required to inform it.

By Megan Duong, CEO and Co-Founder of Plot

  AI, Megan Duong, Opinion Piece, Plot, Thought Leadership