Did anyone else's Google & YouTube app create dozens of international Merchant Center feeds today?

The convergence of large language models, tool-use capabilities, and payment integrations now lets AI agents run multi-step purchase workflows end to end. That means research, comparison, checkout, and payment can happen without a human ever touching your product page. McKinsey's framework describes this as progressive, not sudden. Agents start as research assistants and climb toward fully autonomous purchasing. The staging matters. It means you have a planning window rather than a cliff. Cequence's product launch is the more telling signal. Companies do not build dedicated defense tooling for traffic that does not exist at scale. Agentic bot volume has reached a threshold where it needs its own security category.

In 2026, the infrastructure decisions merchants make (headless commerce, composable architecture, API-first catalogs) will determine who captures agentic demand and who gets bypassed. This is not a peak-season tactic. It is a structural bet on how your catalog gets discovered over the next several years. The comparison being drawn is the desktop-to-mobile transition. Merchants who optimized only for one surface lost share when buyer behavior moved. The same risk applies here, except the "buyer" is now a machine evaluating objective criteria.

If an agent is selecting products, the levers you pull today shift in value. A/B testing landing pages, cart abandonment sequences, and on-site merchandising all assume a human is watching. Agents do not watch. They parse. Four operational questions move to the front of the queue:

• Insight 01Machine readability.Is your product data, pricing, and availability structured for parsing, not just human browsing?
• Insight 02Fraud distinction.Can your systems tell a legitimate shopping agent from a scraping or scalping bot? Right now they often look identical.
• Insight 03Attribution logic.How do affiliate and attribution models function when an agent, not a human clicking an ad, initiates the purchase?
• Insight 04API-first readiness.Agents interact through structured data. Theme-based differentiation does not reach them.

Agentic intermediaries commoditize the buying journey down to price, availability, and spec matching. That is the uncomfortable part for DTC operators. Brand equity and emotional storytelling, the differentiators many brands built their margins on, carry less weight when an algorithm selects on objective terms. If your premium is justified by experience rather than product, an agent may not register that premium at all. It sees a SKU, a price, and a stock signal. Dynamic pricing and reliable inventory data become competitive weapons in a way glossy PDPs are not. The flip side: merchants whose catalogs, pricing APIs, and fulfillment data are agent-accessible can win share from competitors optimizing only for human shoppers. Being readable becomes a moat while others stay invisible. Clean, accurate inventory data and pricing feeds are the foundation of being agent-ready. If your team is manually tracking stock levels or handling reorders without a structured workflow, you are already at a disadvantage. Streamlining how you select products for reorder, calculate quantities based on demand, and coordinate with suppliers makes your data pipeline machine-readable. Click the top right "Get started today" button to explore how this works inside Monocle.

Authentication frameworks for machine-to-machine commerce are still nascent. That creates real exposure. If you cannot verify which agents are legitimate, you either block good demand or expose yourself to fraud and inventory manipulation. Security investment has to scale alongside any agent-facing strategy. Opening structured feeds and APIs without the ability to distinguish good bots from bad ones is a material operational risk, not a growth hack. There is also no proven attribution standard yet. Building agent-facing commerce before you can measure it means flying partly blind on ROI. Verify what you can track before you reallocate budget away from human funnels.

Watch, do not panic. McKinsey's curve is progressive. You are not losing the human funnel this quarter. But the infrastructure you commit to now decides your position when volume compounds. Act now on these:

• Insight 01Audit your product feed for machine readability.Structured, accurate, complete product data is the price of entry for agent visibility. This is a same-quarter task, not a roadmap item.
• Insight 02Pressure-test fraud detection against agentic traffic.Ask your security stack or provider directly whether it can distinguish legitimate shopping agents from scrapers. If the answer is vague, that is your gap.
• Insight 03Evaluate API-first and composable infrastructureif you are still fully theme-dependent. Agent demand flows through structured data, not visual storefronts.

Who benefits most: operators with clean catalogs, dynamic pricing capability, and reliable inventory signals. They are already most of the way to being agent-ready. Who faces the most risk: brands whose entire differentiation lives in on-site UX, visual merchandising, and emotional storytelling. Those touchpoints may be bypassed entirely when an agent shops on a customer's behalf. The strategic reframe is simple. Treat AI agents as a first-class customer segment with their own requirements, not as an edge case in your bot traffic. Merchants who make that shift early set the terms. Everyone else optimizes a funnel their next buyer never enters. Watch how quickly payment and authentication standards for agentic commerce mature. Those frameworks will decide how fast the curve steepens.