There is a version of the agentic commerce story that most companies are telling right now, and it goes something like this: bolt a chatbot onto your storefront, let it answer questions, call it done. Walking the halls of Shopify's Dotdev conference in Toronto, Akohub's CEO and Founder Ray Kung came away convinced that Shopify is telling a different story entirely, one where agentic commerce is not a feature you add but an operating system you rebuild around.
We already covered the conference's headline moments, Vanessa Lee's keynote stats, Sidekick's usage numbers, River's public Slack experiment, and Tobi and Harley's small team thesis, in our first Dotdev recap. This post goes one level deeper into the piece of infrastructure that ties all of it together: UCP, the protocol Shopify is betting the next decade of commerce on, along with the market stakes behind that bet and what "agent ready" actually means as a practical, checkable thing rather than a buzzword.
An Agent Needs Something to Talk To
Most of the industry conversation about AI shopping assumes the hard part is the model. Get a good enough language model, point it at a product page, and the agent will figure out the rest. Shopify's bet is almost the opposite. The hard part is not the intelligence, it is the interface between that intelligence and the actual commerce data underneath it.
That is the entire premise behind UCP, the Universal Commerce Protocol. UCP is a shared language that lets any outside system, a chatbot, a voice assistant, even something as unexpected as a travel app or an astrology app, query a merchant's catalog directly and get back structured, trustworthy product information rather than scraped guesses. Once an agent can see a brand's catalog properly, it can recommend, compare, and route a purchase, with the sale settling back through the platform.
What makes UCP notable is that Shopify did not build it alone or keep it proprietary. The protocol was co developed with Google and announced publicly at the National Retail Federation conference in January, with Etsy, Wayfair, Target, and Walmart involved early on, alongside payment players like Visa, Mastercard, and Stripe. The specification itself is published openly under an Apache license, meaning any platform, not just Shopify's own ecosystem, can build against it. Shopify's own engineering team has described the design in layers, deliberately modeled on how TCP and IP separated responsibilities to let the early internet scale, so that core transaction primitives stay simple while individual capabilities like checkout, orders, and catalog can evolve independently.
On top of that protocol, Shopify built two practical layers merchants actually touch. A Storefront MCP server now runs on every individual Shopify store, letting an agent search that one store's catalog, manage a cart, and walk through checkout with zero setup required from the merchant. A separate Catalog MCP server works at a different altitude entirely, letting an agent search across every participating merchant at once, which is what handles a query like finding a specific product under a certain price with delivery to a certain country. Shopify also opened an Agentic Plan that extends this infrastructure to brands that are not even built on Shopify, letting them register their product data in the Catalog and become discoverable across AI platforms without migrating their storefront.
The implication for merchants is significant. Under this model, a brand does not need to build its own AI assistant or strike individual deals with every AI platform that shows up next year. If the catalog is structured correctly, it becomes visible to whatever agent economy exists tomorrow, almost automatically. Infrastructure, not integration, is doing the work, and the early numbers suggest merchants are already feeling it: Shopify has reported AI referred traffic climbing several times over within a year, with AI attributed orders growing even faster than that traffic itself.
This is the same premise Akohub has built its own signal framework around, just applied one layer down, at the level of an individual merchant's data rather than the platform's. Akohub's weekly insights read a merchant's Shopify data, customers, orders, products, and turn it into a defined signal: what changed, why it matters, and what to do next. Every one of those signals is documented not only by what it tells a merchant, but by whether it would still hold up to an AI agent evaluating that merchant programmatically, whether the underlying data is clean enough, consistent enough, and structured enough to be trusted by something other than a human reading a dashboard. UCP is Shopify solving that problem at the protocol level. Akohub has been solving a version of it at the store level for a while now, which is probably why the framing felt so familiar watching it unveiled on stage.
Full session: Build with Shopify UCP, Universal Commerce Protocol
The Stakes Behind the Bet
It is worth stepping back to see why Shopify is moving this fast. Independent estimates from McKinsey put the total addressable market for agent orchestrated commerce somewhere between three and five trillion dollars globally by 2030, with the United States retail market alone accounting for as much as a trillion dollars of that figure. Shopify's own research found that roughly two thirds of shoppers already say they are likely to use AI in some form when making a purchase, and the company has already activated agentic storefronts by default for millions of its merchants, meaning most of them are discoverable to AI shopping agents whether they have configured anything or not.
That last detail is worth pausing on, because it points to a divide that will matter more than the headline TAM figure. Being technically discoverable to an agent and being genuinely competitive in front of one are not the same thing. A merchant with an internal data team can shape what an agent sees: clean feeds, consistent pricing, well structured customer signals. A merchant running growth decisions with a team of one to three people, which describes a large share of the Shopify ecosystem, usually cannot do that work on their own, even once the infrastructure to be seen is switched on by default. That gap between being reachable by AI and being genuinely ready for it is the space Akohub has spent the last few years building for, on a fairly simple premise: give that small team a weekly read on what actually matters in their own store data, and what to do about it, and the AI facing data hygiene largely takes care of itself as a byproduct. It's a large part of why this conference felt less like a preview of the future and more like confirmation of a direction already underway.
What "Agent Ready" Actually Means
It's easy to treat "agent ready" as a marketing phrase. Underneath the protocol details and the TAM figures, though, it's turning into something much more concrete: a checkable property of a merchant's data and behavior, not a vague aspiration. Clean product and customer data, pricing and fulfillment consistency, repeat purchase signals strong enough for a machine to notice, these are no longer back office hygiene, they're becoming the new front door. An agent evaluating a merchant through UCP or a Catalog MCP query is, in effect, asking the same questions a good analyst would: is this data trustworthy, is it current, does the pattern hold up over time. Merchants who can answer yes to those questions consistently will simply show up more, and more favorably, as this infrastructure spreads. Merchants who can't will be technically discoverable and functionally invisible.
Closing Thought
Shopify is building the protocol layer for that world, open, co developed with Google, and adopted early by some of the largest names in retail. Akohub's work has been focused one level down: reading a merchant's Shopify data every week and turning it into the kind of clean, structured, decision ready signal that both a merchant and an agent can trust. Seeing Shopify converge on the same conclusion, from a very different starting point, is about as clear a piece of external validation as Akohub could ask for. For the fuller picture of the conference itself, including Vanessa Lee's keynote, Sidekick's usage numbers, and Shopify's internal AI collaborator River, see our first Dotdev 2026 recap.
