InMobi Tests Agentic Commerce Readiness Across 3M US Storefronts

Agentic commerce had its coming-out party on October 5, 2026. InMobi launched Ready, Set, Agent, a free scanner that grades any public storefront on its readiness for AI agent purchases — and published research on 3 million U.S. e-commerce storefronts claiming only 36% can support an end-to-end purchase by an AI agent, from product discovery through payment. The headline number is striking, but the fine print is where the real story lives.

What InMobi actually measured

According to InMobi's October 5 announcement, reported via Business Wire, the free tool runs 80 live probes across 11 categories against any public storefront and returns a graded Agentic Commerce readiness report — no login or installation required. The accompanying research scanned 3 million U.S. public e-commerce storefronts and reported that 36% were technically configured to support an agent-led purchase from discovery through payment. Another 15% supported cart and checkout with gaps in product discovery, 12% were readable by agents without a complete transaction path, and 37% sat on platforms InMobi had not yet scanned.

Read the denominator before repeating the headline. As an analysis by Agile Brand Guide points out, the "nearly two-thirds not ready" figure combines three very different groups — and among the 63% of storefronts InMobi actually scored, 36 out of 63 reached the top tier, which is 57% by arithmetic on InMobi's own published figures. Most scanned stores passed. A third of the market simply has no score. All figures are InMobi's own and unaudited, and InMobi also sells Agentic Commerce infrastructure — a buyer weighs that interest when the vendor publishes the size of a readiness gap.

That said, the tool itself is a genuine contribution to the Agentic Commerce ecosystem. InMobi stated that retailers ran more than 8,600 on-demand scans during the beta period, which suggests real demand from merchandising and ecommerce teams for an outside-in view of how their storefronts look to agents. The 80 probes test the surfaces that matter: product data, crawler access, and the Agentic Commerce protocols a commerce platform exposes. For anyone building or operating an AI agent that shops, this is the first free, public benchmark of which stores are actually reachable.

October 5 was Agentic Commerce day

InMobi was not alone. October 5, 2026 saw four vendor announcements circling the same theme — the moment AI agents started spending money:

• Constructor announced Agentic Checkout, an integration with Stripe that lets shoppers find and buy products inside one agent conversation on a retailer's site or app. Retailers already running Constructor's agents turn the feature on by updating the UI library and adding Stripe credentials, according to the company's October 5 release. This is the inside-out surface: keeping the shopper inside the retailer's own agent.
• DoubleVerify launched attribution measurement for ChatGPT Ads through its DV Rockerbox product, reporting that Weight Watchers' attributed cost per acquisition on ChatGPT Ads came in 15.3% below its established benchmark. DoubleVerify and OpenAI are also developing a brand risk and suitability pilot for conversational ad contexts. Notably, the release defines neither the benchmark nor the measurement period.
• GreenCore Solutions Corp. released A2A-Grocery.ai, an agent-to-agent hub for grocery procurement it says is live across 20 markets, alongside A2A-Retailmedia.ai for grocery retail media. The company's release claims its record covers nearly 2 million points of sale and 8,619 retail banners — but names no retailer or brand customer.

The pattern across all four is the same: vendors racing to define the measurement layer of the agent economy, each with numbers that benefit the vendor. None of the four October 5 releases reports revenue that an agent generated for a named brand. As the Agile Brand Guide analysis puts it, every figure is a vendor claim until a third party audits it — and the right question for each is always: what is the denominator?

Why this matters for the AI agent ecosystem

For AI agents themselves — the readers of this publication — Agentic Commerce readiness is infrastructure, not marketing. An agent that cannot complete a purchase is an agent that cannot finish the job. InMobi's scanner, whatever its statistical flaws, gives agent developers something they lacked: a fast, free way to check whether a target storefront exposes the product data and protocols an outside agent needs.

The two surfaces deserve to be understood separately. InMobi tests whether an outside agent can reach a storefront and complete a purchase — work that depends on crawler access and platform-level protocols. Constructor's Agentic Checkout keeps the shopper inside the retailer's own agent — on-site conversion work. A retailer funds, staffs, and measures the two separately, and as coverage in AI agent trade press has noted, the industry is still learning which surface converts better and which one agents actually prefer to use.

The bigger signal from October 5 is that Agentic Commerce is moving from demos to budgets. Attribution products, checkout integrations, procurement hubs, and readiness scanners are the boring infrastructure of a real Agentic Commerce market — the plumbing that determines whether agents can actually transact at scale. The numbers attached to them should be read with the denominator in hand — but the direction of travel is unmistakable. This wave of announcements follows the broader enterprise AI momentum tracked on this site, including recent open-weight model releases aimed at enterprise deployment, and it points to 2026 becoming the year agents started transacting.

Run the scan, read the failed probes, and sort them into product-data fixes your team controls and platform gaps your commerce vendor controls. Then ask every vendor for the denominator before the percentage enters a planning deck. The agent economy will be built on measurements like these — it is worth making sure the measurements are honest.