The dark secret of Meta's new AI agent -
The revolution of Meta's $META Muse agent doesn't just threaten search engines – it could harm one of the internet's most powerful revenue models:
making sellers pay to influence what the consumer sees.
And this is already a much bigger story...
1️⃣DoorDash - $DASH - sponsored ranking loses meaning
DoorDash has about $1 billion in advertising revenue, approximately 9% of total revenue.
The model is simple: restaurants and businesses pay to get better exposure within the platform.
But if I tell Muse "order me dinner," I don't need to browse through restaurants, ratings, and advertisements. The agent can perform the search, compare options, and order for me.
DoorDash still delivers the food – but if the user no longer sees the digital shelf, who cares who bought the top spot?
2️⃣Amazon $AMZN - who needs sponsored results?
Amazon generates about $68.6 billion from advertising revenue, approximately 10% of total revenue. This is one of the company's enormous profit drivers.
But an AI agent changes the game.
If Muse searches for a product for me on Amazon, it can focus on price, quality, reviews, availability, and specifications – not on who paid Amazon to appear first.
This means the interface becomes less important, and the agent's algorithm becomes the gatekeeper.
Amazon quickly realized where things were headed – and rushed to block Meta's agent from its platform.
3️⃣Instacart $CART - when the digital shelf disappears?
Instacart generates about $1.06 billion from advertising, approximately 28% of total revenue.
Brands pay for their products to get better placement while the consumer browses.
But what happens when the consumer no longer browses?
They simply say: "I need these products – build me a cart."
The Muse agent can turn the entire experience of the shelf, search, and sponsored products into an automated process behind the scenes.
No scrolling – no shelf – and not much meaning for sponsored placement.
4️⃣Yelp $YELP - perhaps the clearest casualty
About $1.39 billion, approximately 95% of its revenue comes from advertising.
And this makes the scenario particularly interesting.
Because instead of "search for a plumber on the Yelp platform, compare ratings, open several profiles, and check advertisements"
The user can simply say "find me the best plumber available tomorrow and book them."
Suddenly, a large part of the interface's value disappears.
The customer doesn't need to see all the options. They need an answer.
5️⃣eBay $EBAY - the inventory remains, but who controls the choice?
eBay generates about $2 billion from advertising revenue, approximately 18% of total revenue.
The Muse agent can use eBay's vast inventory and still bypass the sponsored exposure mechanism.
The seller can pay to be first within eBay.
But if the agent is the one deciding which product to show the user, it doesn't have to play by eBay's auction rules.
Understand something important:
The story here is not "Muse kills Amazon" or "Muse kills Yelp".
The story is much deeper.
A platform like Muse separates the transaction platform from the choice interface.
In the old internet model, the company controls three things: it holds the inventory or service, it controls the interface, and it determines who appears before the user – including those who pay for it.
In the world of AI agents – these three things can split apart.
The agent can take inventory from Amazon, a business from Yelp, a restaurant from DoorDash, and a product from eBay – and then make the choice for the user.
And this creates an enormous economic problem:
If the user no longer sees the shelf – it's much harder to sell them a spot on the shelf.
This is the real threat of Muse. 🤖
No longer a search engine, no longer an application, but a new layer of mediation between the consumer and an entire economy that built its business around their attention.
In the search era, they paid for you to see them – in the agent era, they will need to convince the machine to choose them.
The apparent solution must be customized agents representing the platform itself – for example, an Amazon agent – the problem is that in such a case, each platform will present its own model – and the agent simply loses all its value as a personal assistant...
We are monitoring 🔎