People keep asking me when agent commerce will arrive, as if it will announce itself. It will not. The first agent transactions will not look like commerce at all. They will look like API calls that happen to carry value, logged in systems nobody reads, initiated by software doing exactly what it was configured to do.
This is how infrastructure shifts always happen. Nobody experienced the moment retail moved online as a moment. There was no day when email replaced the memo. The change arrived as a series of individually boring decisions, each one too small to feel like history, and then one year the old way was the exception. The agent economy will follow the same script, and I think it is worth spelling out why.
The First Purchases Are Already Disguised
Consider what an autonomous agent needs in the course of doing its job. More compute when a task spikes. A dataset it does not have. A specialized model for one step of a pipeline. An answer from another agent that has capabilities it lacks. Today, most of these needs are met inside the boundary of whoever deployed the agent: pre-purchased quotas, existing subscriptions, keys provisioned by a human in advance.
Every one of those pre-arrangements is a small tax on capability. The agent can only use what someone predicted it would need. The obvious next step, obvious enough that many teams are quietly taking it, is to let the agent acquire what it needs at the moment it needs it. Metered access. Pay-per-call. A budget instead of a shopping list.
The day that happens, the agent is buying. Not browsing, not adding to cart, not checking out. Just a request going out with authorization to spend attached, and a resource coming back. To an observer reading the logs, it is indistinguishable from any other API call. That is exactly the point. The transaction has dissolved into the plumbing.
Why Nobody Will See It Coming
Three properties make this shift invisible until it is finished.
First, the transactions are small. A fraction of a cent for an inference, a few cents for a dataset slice. Small transactions do not trip anyone's attention. They accumulate in expense lines that read as cloud costs, which is technically what they are.
Second, the transactions are fast and frequent, which means the interesting number is not any single purchase but the aggregate flow, and aggregate flows only become visible in retrospect, when someone finally graphs them.
Third, and this is the one I find most consequential, there is no user experience to notice. Human commerce shifts were at least visible as new behaviors: people staring at phones, boxes on doorsteps. Agent commerce has no storefront, no checkout page, no unboxing. The entire economy runs in the space between two machines agreeing on a price. Absence of spectacle is not absence of scale.
My years in enterprise security trained me to pay attention to precisely this kind of quiet accumulation. The threats that mattered were never the loud ones. They were the ones that looked like normal traffic until you understood what the traffic meant. Agent commerce will look like normal traffic. It already does.
What Matters Before the Noticing
Here is why I think the invisibility matters and is not just a curiosity. The rules of an economy get set early, by whatever infrastructure the first participants happen to use. By the time a shift is visible enough to debate, its defaults are already load-bearing. We are living with decades of consequences from early defaults on the internet, where identity was an afterthought and security was a retrofit. I spent a career on the cleanup crew for those defaults. I have no appetite for watching the sequel.
So the time to get the foundations right, verifiable identity for software that spends, settlement conditioned on proof rather than promises, spending authority with boundaries a counterparty can check, is now, while the volume is still small and the patterns are still soft. This conviction is most of the reason Setix exists.
When people finally notice that agents are buying things, the noticing will not be the story. The story will be everything that got decided while nobody was looking. I would like more of us deciding it deliberately.