Checkout.com just predicted that AI agents could handle 21% of household spending within five years.
Today, that number is essentially zero.
And it is zero for a simple reason: almost all of the AI in commerce right now stops short of the actual transaction. It helps people find products, compare them, or build a cart. But people still click the button. They approve the order. They swipe the card.
AI is influencing the path to purchase, but it is not the one completing it.
So the interesting question is not whether 21% is right. The interesting question is what would have to change for AI to move from helper to decision maker?
A good place to look is the part of household spending most people never think about.
60-70% of all purchases are under $25.
They barely move the budget, but they take a surprising amount of attention.
Buying deodorant. Reordering paper towels. Choosing snacks.
Each one takes three to five minutes once you include searching, comparing, and checking out.
Spread that across a month and you end up losing hours to decisions that do not matter very much.
If AI agents simply handled that category, two things would happen.
You would get back real time.
And you would likely stop overspending on convenience markups and impulse adds.
Saving 5-10% on the highest frequency part of household spend is not a stretch. It is just the natural outcome of reducing friction and removing impulse driven choices.
This feels like the real on ramp for agentic commerce.
Not big purchases.
The tiny ones that drain attention and quietly drain money.
But reaching 21% of spend requires more than efficiency. It requires trust.
Trust in the rails.
Trust in your agent.
Trust that recommendations are not shaped by hidden incentives.
And the confidence that it is easy to correct the system when it gets something wrong.
If you extend the thought experiment and ask what it would take to reach 50%, the picture changes pretty dramatically.
Delegation would have to feel normal.
Small decisions would fade into the background.
Agents would think in outcomes instead of SKUs.
I am not predicting 50% but those are the mechanics that would need to be in place.
For brands and platforms, that is the useful part of this conversation.
Are you preparing for a 5%, 21% or a 50% world?