From pages to protocols, Laughlin Rigby’s advice for marketers in the AI agent economy
Artificial intelligence (AI) is shifting from answer engines to action engines. The next growth battleground will be won by brands whose facts are machine-legible, whose flows are agent-callable, and whose infrastructure supports AI-enabled purchases. A PwC survey found that investment in Ireland in AI agents is accelerating, with 70 per cent of companies increasing AI-related budget due to interest in agentic AI in the year ahead.
On the retail side, OpenAI’s ChatGPT supports Instant Checkout for U.S. users’ single-item purchases, enabled by Stripe and the Agentic Commerce Protocol (ACP), with expansion flagged as the next step. In parallel, Google is compressing the path to purchase in Search. In AI Mode, it has introduced agentic task features like restaurant reservations (via partners such as OpenTable, Resy, and Tock).
In January 2026, Google announced the Universal Commerce Protocol (UCP), a new open standard intended to let agents and commerce systems interoperate across the full shopping journey (discovery → buy → post-purchase support). Google also says UCP will soon power a new checkout feature on eligible product listings in AI Mode in search and the Gemini app, enabling shoppers to check out from eligible US retailers during research using Google Pay with details saved in Google Wallet, with PayPal support “soon”.
Similarweb reported that AI platforms generated over 1.13 billion referral visits last June, up 357 per cent year-on-year. Assistants will increasingly shortlist, recommend, and transact for users. Brands that surface rich structured data (complete product facts, availability, policies), expose deep-link or API-based actions, and support trusted payment flows will be favoured by agents.
Rather than focus solely on campaigns, conversion, or traffic, the core investment moves to data readiness and endpoint readiness. That shift is durable for the next 12+ months, irrespective of which assistant or surface your customer uses. Reframe “SEO + CRO” as AI surface optimisation: make your catalogue machine-legible, your flows callable, and your payments agent-compatible.
Once you build for one surface (like ChatGPT Instant Checkout), you are future-ready for the next (such as Google’s agentic checkout surfaces, or travel bookings as protocols expand).
A few pointers on how to get there:
- Pair JSON-LD (product and offer) with agent-ready checkout endpoints (like ACP) so assistants can both trust your facts and complete the sale.
- Mirror shipping/returns/warranty in visible copy and structured data so assistants can quote policies correctly.
- Create deep links that pre-fill basket/booking parameters (size/colour, dates, party size) and map them to your checkout/booking steps.
- Add a quarterly agent-readiness audit: validate schema, simulate tasks in Atlas (OpenAI’s agent browser) and AI Mode, and test commerce flows end-to-end.
- Keep robots.txt policy explicit for AI crawlers you want (or don’t want) reading your site.
- Instrument clear attribution for AI traffic so you can see what actually converts.
Two further developments strengthen the “pages → protocols” thesis:
- Branded agents are becoming a new storefront. Google’s newly announced Business Agent is positioned as a retailer’s virtual sales associate inside search, able to answer questions “in a brand’s voice” — with plans to enable direct purchases (including agentic checkout) within that experience.
- Discovery data is getting more conversational. Google is rolling out new Merchant Center attributes aimed at conversational commerce (such as answers to common product questions, compatible accessories, substitutes), effectively formalising what “machine-legible merchandising” looks like.
Agent-mediated commerce is no longer the distant future; it’s here for retail
In the near term (12–18 months), assistants will triage options, quote product facts, and initiate transactions especially in retail. For travel/tours/experiences, the same principles apply, but it may require slot/availability endpoints and hybrid flows until agentic checkout support becomes more standardised across booking systems.
Practically, this means:
- Authority & clarity: publish consistent structured data (product/offer) for retail; for tours/slots, use event/offer and descriptive schema (itinerary, capacity, timings).
- Action endpoints: for retail, deep links or ACP checkout; for bookings, pre-filled booking links or API endpoints for availability and holds.
- Policy signals: machine-readable returns, cancellations, delivery/booking windows.
- Attribution hygiene: tag assistant-surface referrals and map to revenue.
- Crawler governance: define AI-crawler permissions so agents can parse your catalogue or booking system correctly.
Build once for agentic commerce, and your business will already be architected to scale into tours/experiences as these standards expand. Early readiness now gives brands an advantage: make facts machine-readable, make actions safely callable, and let assistants do the rest.
Agent-mediated commerce is no longer the distant future; it’s here for retail, and fast approaching for travel and experiences. Brands that build machine-readable catalogues, callable actions, and trusted payments now will be the ones chosen and transacted with across whichever assistant your customer uses.
Laughlin Rigby is the digital transformation and AI director of Core Optimisation
This article featured on the company’s The Future of eCommerce Marketing ebook
