The businesses that will be ready before anyone asks them to be.
Being ready for AI agents isn't a deadline to hit when the volume arrives. An AI agent already checks the same facts today that it will at scale next year, just less often. The businesses ahead of the curve are doing the unglamorous work now, before anyone's asking them to.
6 min read
Agents already check these facts today
A family-run cookware retailer, three shops and a website, spent this spring on something nobody would call exciting: making sure its returns policy read as one consistent, checkable fact everywhere it appeared, instead of three slightly different versions across the website, the marketplace listings and the customer-service scripts. No agent was asking for that fact at any real volume yet. By the time one does, it'll already be true.
It's easy to treat agent-readiness as something to prepare for once the traffic shows up, the way a business might scale up staffing ahead of a known busy season. That's the wrong model. An agent checking a retailer's returns policy today, even if it's one request among a handful, needs the same accurate, consistent fact an agent checking it next year at far higher volume will need. Getting it right now doesn't cost more than getting it right later. It just means it's already done.
The work itself is boring, and that's the point
None of what the cookware retailer did involved AI at all. It was a spreadsheet cross-check: does the returns window say the same number of days on the website as it does in the marketplace feed, does the customer-service script match either of them. The gap wasn't a technology problem, it was three teams updating three different places at three different times over a couple of years. Fixing it was closer to bookkeeping than to anything that would show up in a roadmap about "AI strategy."
The same pattern holds well beyond returns policy. A retailer that keeps its sizing notes, its material specs, its delivery estimates consistent across every place they appear isn't doing AI preparation particularly, it's doing the kind of operational housekeeping that was worth doing anyway, and it happens to be exactly what an agent checks first.
Where this shows up for a business
A fact only has to be right in one place to be right everywhere it's checked, once a retailer's returns policy, sizing notes and delivery terms are connected to a single source instead of copied by hand across three teams. That's the connection Selfe makes: a retailer's own systems, once, so nobody has to re-check the fact by hand every time it changes somewhere else.
How do we know if we're already behind on this?
Check whether your returns window, sizing notes or delivery estimates say the same thing in every place a customer or an agent might find them. If they don't, that's the gap.
Is this really worth doing before agents are sending us meaningful traffic?
Yes, because the fix doesn't get easier or cheaper by waiting, and the fact needs to be right regardless of who's checking it.
The next twelve months in agentic commerce.
Most writing on where AI agents are heading falls into one of two camps: everything changes overnight, or nothing really changes at all. Neither is honest. Some things about agentic commerce are already settled enough to build for. Others genuinely aren't decided yet, and pretending otherwise doesn't help anyone plan.
The standards still being written, and why that's not a reason to wait.
The specifications an AI agent uses to connect to a business, prove who it is, and get authorised to spend are all still being actively revised by the organisations building them. That's genuinely unsettled, not a reason to wait. The work that needs doing doesn't depend on any of it finishing first.
What changes for discovery in the next twelve months.
Not a sweeping prediction. Over the next year, the practical change for discovery is narrower and more useful than "everything will be different": more of the requests reaching a business will already carry real constraints attached, rather than arriving as a vague, browsable query.