The visit is already a channel. Agent checkout is still small.

AI assistants are already sending shoppers to businesses, and those shoppers convert well, but software agents completing purchases on their own are still rare. So the work worth doing now is the visit: making sure an assistant can find you, describe you correctly and hand a ready-to-buy customer to the checkout you already have.

The venture firm a16z argues in "Welcome to the B2A Economy" (Nov. 10, 2026) that businesses "will have to fight for bots' business." a16z invests in this trend, so its essay is a useful map of the idea, not proof of it. The independent numbers tell a more measured story.

The growth figures a16z quotes come from Shopify. On its Q1 2026 earnings call (May 5, 2026), Shopify said AI-driven traffic to its stores grew 8x year over year and orders from AI-powered searches grew nearly 13x. On its Q3 2026 call (Nov. 3, 2026), Shopify reported AI traffic up 2.6x and orders up 3.3x year over year. In the same breath, it said of agentic shopping that "the volume is still small."

Adobe's report of Oct. 29, 2026 shows why the visit matters. In September 2026, traffic from AI sources to U.S. retail sites grew 71 percent year over year, converted 90 percent better than non-AI traffic and generated 83 percent more revenue per visit. Adobe measured people clicking a link from an AI tool, not agents buying alone.

Meanwhile, Semafor reported on Nov. 10, 2026 that DoorDash had received "maybe several hundred orders in total" from personal AI assistants, and Airbnb's CEO described "very, very low numbers" of agent bookings.

A smiling woman in a red apron uses a tablet in a store, with a colleague at the shelves behind her.

Discovery: be found, and be described accurately.

Before an assistant can recommend you, it has to read what you sell, who you sell it to and on what terms. Adobe found that in September 2026 the average U.S. retail homepage scored 63 percent on its Citation Readability Score, which means more than a third of that content was not visible to the AI systems sending shoppers to retailers (Adobe, Oct. 29, 2026).

Adobe also found that navigation-heavy, promotional layouts score worst. General merchandise homepages scored 40 percent, while electronics led at 79 percent, helped by pages full of setup guides and detailed specs.

The fix is plain writing. Put a short, factual description near the top of your homepage and your main category pages: what you sell, the kind of customer you serve, where you deliver or operate, and your key terms. Use the same wording everywhere your business is described.

Two live NTx guides cover the profiles assistants draw on: Businesses in ChatGPT: How to Claim Your Business Profile and What It Does and Google AI Overviews and Answer Engine Optimization. The next step is covered in our upcoming article "You Claimed Your ChatGPT Listing. Now Make It Describe You Correctly."

Evaluation: publish facts an assistant can compare.

Once an assistant has found you, it compares you with the alternatives, and it can only compare facts it can read. Publish price, availability, specs, shipping and the return policy as consistent fields on every product page. On its Q3 2026 call, Shopify said searches powered by its product catalog converted "at 2x the rate of general AI search."

That is Shopify describing its own product, so read it as a vendor figure. The reason it gave applies to any business, though: the catalog turns product details such as size, color and material into values an agent can read, with current pricing and inventory. You don't need Shopify to apply the lesson, just the same facts in the same format on every page:

  • Price, including what's in it and what isn't.
  • Availability, kept current, so an assistant doesn't recommend something you can't ship.
  • Specs as labeled values (size, material, compatibility), not buried in marketing copy.
  • Shipping: where you ship, how fast and at what cost.
  • Returns: the window, the conditions and who pays.

Adobe's findings point the same way. The pages AI reads best are the ones that answer shoppers' questions, "from product details and buying guides to FAQs and return policies" (Adobe, Oct. 29, 2026). If a policy lives only in a PDF or a pop-up, assume an assistant can't see it.

Checkout: admit the shopper's agent, and keep your own checkout.

Training crawlers and user-directed agents are different visitors, so give them separate settings. Then keep the sale on the checkout you already run, because completed agent orders are still rare.

A training crawler copies your content to build a model. A user-directed agent is visiting on behalf of a specific person who may be about to buy. On Sept. 15, 2026, Cloudflare replaced its single "Block AI Bots" switch with separate controls for Search, Training and Agent traffic. Its press release the same day said the recommended setup for sites that carry advertising blocks AI agents on ad-carrying pages. If your product pages run ads and you accepted that recommendation, you may be turning away shoppers' agents without knowing it. Sites that had switched on the old "Block AI Bots" setting were moved to the same agent setting, blocking on pages with ads.

Cloudflare's AI Crawler and Agent access policies documentation now lets site owners set an overall Agents policy and then adjust it for individual personal agents. On those personal agents, it is direct: "Cloudflare recommends Allow (do not block)." You can still refuse AI training on your content without shutting out a customer's assistant. Our article on shared AI training control for websites explains the training side.

Big retailers are split. Semafor reported on Oct. 16, 2026 that Amazon and eBay had blocked agents to protect their ad businesses and checkout experiences, and, citing Morning Consult, that few consumers are comfortable letting AI make purchases on their behalf. a16z argues the opposite: for years, retailers "built walls to keep bots out," and now "they will have to fight for bots' business." For a privately held business, the practical middle ground is to let the shopper's agent in to read and compare, and let the purchase finish on your own checkout. Don't build a separate agent checkout yet.

What to fix this quarter.

Correct how you are described, publish your product and policy facts, and check your bot settings. Then track AI-referred visits separately in your analytics, and add an agent-only checkout only once those orders show up in your own reports.

  1. Fix the description. Rewrite your homepage and category pages so a stranger, or an assistant, can say in one sentence what you sell and to whom. Claim your business profiles.
  2. Publish the facts. Make price, availability, specs, shipping and returns consistent, readable fields on every product page, the kind of structured data Shopify credited on its Q3 2026 call.
  3. Check the bot settings. Ask whoever runs your website or CDN, whether a remote staff member or an outside agency, to confirm in writing that training, search and agent traffic have separate settings, and that personal agents aren't blocked on your product pages. If you use Cloudflare, these are the controls it introduced on Sept. 15, 2026.
  4. Measure AI referrals separately. Have your remote team set up a channel or segment in your analytics for visits that come from AI assistants such as ChatGPT, Gemini, Copilot in Microsoft 365 and Perplexity, and track their conversion rate against your other channels. That gives you your own evidence instead of someone else's averages.
  5. Wait on agent checkout. Semafor's Nov. 10, 2026 reporting and Shopify's own "the volume is still small" both say the same thing: agent purchases haven't arrived at scale. When agent-completed orders appear in your reports in meaningful numbers, that is the moment to invest in a dedicated agent checkout.

None of this needs new software. Most of it is writing, settings and one analytics report. If you want to see which AI tools businesses like yours already run, start with What AI Tools Are Small and Mid-Sized Businesses Actually Using?.

We can help you with that.

NTx AI Consulting works with owner-led and partner-led privately held businesses that are currently using AI tools in everyday workflows, or would like to be. The complimentary consultation is a working session, not a product demo. If a deeper engagement makes sense, Phase 1 is an AI Disruption Readiness Audit. You can learn more at https://ntx.ventures/ai-consulting/.

Sources

  1. "Welcome to the B2A Economy". https://a16z.com/welcome-to-the-b2a-economy/
  2. Q1 2026 earnings call. https://www.fool.com/earnings/call-transcripts/2026/05/05/shopify-shop-q1-2026-earnings-transcript/
  3. Q3 2026 call. https://www.fool.com/earnings/call-transcripts/2026/11/03/shopify-shop-q3-2026-earnings-call-transcript/
  4. Adobe's report of Oct. 29, 2026. https://business.adobe.com/blog/ai-traffic-higher-revenue-per-visit-retail-holiday
  5. Semafor reported on Nov. 10, 2026. https://www.semafor.com/article/11/10/2026/agentic-commerce-is-slow-to-catch-on
  6. Sept. 15, 2026. https://blog.cloudflare.com/accountable-mixed-use-ai-crawlers/
  7. press release. https://www.cloudflare.com/press/press-releases/2026/cloudflare-helps-end-the-search-or-ai-training-tradeoff/
  8. AI Crawler and Agent access policies documentation. https://developers.cloudflare.com/bots/additional-configurations/block-ai-bots/
  9. Semafor reported on Oct. 16, 2026. https://www.semafor.com/article/10/16/2026/ecommerce-confronts-ai-shopping-agents-ahead-of-holiday-season