Remember the Cinnamon Challenge, when people filmed themselves trying to swallow a spoonful of ground cinnamon in under a minute?
For a short window in the early 2010s, it was everywhere – until it was gone. Now, it’s mostly remembered as one of those “What were we doing?” internet moments. And don’t get me started on flash mobs.
Creators emerged around the same time, and you might have thought they would fade as well. Everyone had access to a camera and an upload button, so it was easy to imagine that the idea of a lasting creator class would evaporate. But that is not what happened. The creator economy is now a $30B+ force that sits at the center of modern marketing strategy.
The Cinnamon Challenge and the rise of creators both felt big. One became fodder for cultural trivia. The other became infrastructure.
AI vendors today are at a similar juncture. Almost everything feels big. The question is: Are you looking at the next Gangnam Style or the birth of a new industry?
Here are four signals you can use to assess whether an AI vendor is likely to matter in two years.
1. Product-led or sales-led?
When I think about vendors in our industry that have endured versus those that quietly vanished, one pattern stands out. The durable ones were product-led.
They did not rush to build a sales team before the product was excellent. They were obsessive about getting feedback from customers, and they were disciplined about building what would be valuable for the most people rather than chasing every one-off feature that could close a deal.
They also spent a surprising amount of time educating customers about the broader space. The vendors that faded did the opposite. They pushed their own product, talked less about the ecosystem and sometimes even tried to hide alternatives.
You can notice this difference very quickly in an introductory meeting with a vendor. If most of the time is spent understanding your world and explaining the landscape, you are likely talking to a product-led team. If most of the time is spent running a demo script and pressing you for next steps, you probably are not.
2. Could you build it with an LLM, an automator or a vibe-coding platform?
A very practical test is to ask yourself how hard it would be to rebuild what you are seeing with a popular large language model, a general automation platform like Zapier or n8n or even a vibe-coding app builder like Lovable or Base44.
Many marketers have not built anything in any of those tools today, which is understandable. That is exactly why every senior leader should spend an hour trying to build something simple with them, or sit side by side with someone who can. Once you have seen what is possible with a few prompts and some basic workflow steps, your filter sharpens overnight.
If you look at a product and realize that you could recreate it, the underlying moat is shallow. Those vendors can still be useful experiments, but they are less likely to be central to your stack two years from now.
3. Depth of integration is a proxy for seriousness
Every vendor will tell you they can push an ad into Google and Meta or publish to Adobe. This only tells you they are trying to capture revenue quickly rather than trying to build a long-term viable product. Within media planning and activation, a vendor that understands your world will be thinking about search, social, programmatic, retail media and direct partnerships.
This does not mean you should demand that a young company has everything built on day one. Instead, listen carefully to how they talk about the rest of the stack. Do they understand how campaigns flow through your systems? Do they have a coherent idea of what they will integrate with and why?
4. Surface delight versus hard, invisible work
Many AI products are built around surface-level delight moments. You click a button, and a paragraph appears or a few charts animate. Suddenly it feels like something magical just happened. On the back end, very little real work has changed.
The right signal is to ask what hard, invisible work the tool is doing on your behalf. Is it eliminating hours of reconciliation or QA? Is it improving the quality of decisions in a way you can describe? Is it simplifying a messy process across teams?
The vendors that will still matter in two years are the ones that make a meaningful dent in operational reality, even if the interface is less flashy.
In an environment where almost every AI product feels urgent, these four signals will help you find a partner for the long haul, not just one that represents temporary hype. That way, you can identify truly foundational technology to build on, rather than chasing every “six sevvvvvven” that comes along.
“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Follow Jay Friedman and AdExchanger on LinkedIn.
For more articles featuring Jay Friedman, click here.
