Agentic commerce doesn't eliminate marketplaces — the battle shifts to the conversational interface

Summary

Agentic commerce doesn't "eliminate" marketplaces: it shifts value toward conversational interfaces driven by AI agents (ChatGPT, Gemini AI Mode, Perplexity). Travel, restaurant and retail platforms are being restructured around three possible roles: being disintermediated on simple segments, becoming "platforms-as-API" behind third-party agents, or repositioning as "platforms-as-agents". For merchants and service providers, competition shifts toward visibility in the agentic interface, real-time data quality, and execution protocols. Busony supports this transition with its agentic optimization offering.

Thesis: when agents displace the "front door"

Agentic commerce marks the shift from AI that advises to AI that plans, calls tools/APIs, chains steps and acts — booking, paying, modifying, canceling — with variable user control.

The fundamental shift: the front door (where intent forms, comparisons happen, and decisions are made) migrates to conversational interfaces.

Four converging forces are accelerating this movement in 2026:

1. Agents capable of acting on the web (navigation, forms, UI interactions) — bypassing the lack of APIs at some players 2. In-chat apps/connectors — the conversational interface becomes a platform where services plug in at the right moment in the dialogue 3. Agentic checkout (ACP/UCP protocols + tokens) — moving from recommendation to transaction 4. Multi-platform standardization (UCP, MCP/A2A compatibility) to reduce N×N integration costs

```text Before (web/search): Google → Booking → hotel After (agentic): Agent interface (ChatGPT/Gemini) → (API/app) → Booking/OTA → hotel Direct variant: Agent interface → Hotel API → payment → service ```

Impact on marketplaces: three possible roles

Agentic commerce acts as a "journey compression". Instead of a funnel (search → comparison sites → checkout), the user delegates an intent and waits for an actionable result.

Agentic interfaces (ChatGPT, Gemini, Perplexity)

They become the UI where intent forms and where the agent orchestrates apps/APIs. Their challenge: trust, liability, and checkout integration. Winning strategy: apps platform + protocols + guardrails (confirmations, consent).

OTAs and travel marketplaces (Booking.com, Expedia)

The "top of funnel" migrates to agents; value shifts toward execution (modifications, customer service, compliance). In early March 2026, Reuters reported that OpenAI had scaled back its ambition to "book directly within ChatGPT" for travel, favoring checkouts via third-party apps — reducing the immediate disintermediation risk for OTAs.

Winning strategy: becoming "API-first" + "agent-first" (apps, real-time pricing, post-purchase).

Restaurant platforms (OpenTable, TheFork)

Google deployed agentic booking capabilities via partners: the agent finds and redirects to the reservation without becoming the booking platform itself. Restaurants are often the segment where agentic execution arrives first, because the workflow is shorter and less regulated.

GDS and travel execution layer (Sabre, Amadeus)

Agents need a "translation" of travel language + real-time inventory. Winning strategy: positioning as "execution layer" for agents + MCP/agentic APIs.

E-commerce and middleware (WooCommerce, Mirakl)

WooCommerce already exposes capabilities via MCP (product/order management). Mirakl formalizes the "orchestration layer" role between existing stacks and agentic channels. To optimize your store, see our WooCommerce AI page.

Three evolution scenarios

All three scenarios coexist; the question isn't "which one wins" but where (segment) and when (maturity).

Scenario 1: Full disintermediation

The user (or their agent) goes from intent to transaction without a marketplace: direct purchase, direct booking, or action via browser agent. Checkout protocols (ACP) accelerate this path.

High probability for "simple fulfillment" goods (standard retail), lower for complex travel.

Scenario 2: Platforms-as-API

Marketplaces become the execution infrastructure: the agent orchestrates, but booking/payment/after-sales relies on the platform via app/API. This is the direction of "apps in ChatGPT" announcements.

Very high probability for travel/hospitality within 12–24 months, compatible with operational reality.

Scenario 3: Platforms-as-agents

Marketplaces create their own agent: "connected trip" assistant, service agent, B2B agent, support agent.

High probability for leaders (investment capacity + data); more uncertain for challengers.

Differences between restaurants, hotels and complex travel

The probability of disintermediation depends on the level of complexity and cost of error.

Being agent-ready: technical requirements

Being agent-ready isn't about "adding a chatbot". It means making your offer legible, verifiable and actionable by agents, with real-time data and execution primitives.

Our Agentic AI Optimization page details the concrete steps to reach this level.

The MCP, ACP and UCP standards

• MCP: "tools & data" connection standard (apps/agents ↔ external systems). OpenAI built its ChatGPT apps on it. WooCommerce exposes capabilities via a dedicated MCP endpoint.
• ACP: agentic commerce protocol led by Stripe + OpenAI to standardize transaction flows.
• UCP: Google's open-source standard to unify commerce integration on AI Mode/Gemini, compatible with MCP/A2A, keeping the merchant as "Merchant of Record".

"Minimum viable agent-ready" checklist

For WordPress and WooCommerce sites, Busony integrates these standards into its development process.

Strategic recommendations by actor type

For marketplaces (Booking, Expedia, TheFork, Airbnb)

The pivot to nail: don't define yourself by the historical UI, but by execution and trust. The right question is no longer "how to bring the user to my site", but "how to be the agent's preferred executor".

1. Become "app + API" before "final agent": most robust strategy against interface pivots 2. Invest in the post-purchase layer (change, cancel, support): this is travel's moat 3. Win the agentic ranking war: instrumentation + data quality 4. Build "service-first" agents (support, rebooking, concierge): more defensible than "booking-everything"

For hotels

• Reinforce book-direct + loyalty to reduce OTA dependency
• Expose actionable capabilities (availability, rates, policies, cancel) via APIs/partners
• Clean up data: schema.org, content, cancellation policies

For restaurants

• Be "bookable" everywhere: correct hours, menu, policies + reliable inventory
• Google AI Mode books via partners: being correctly exposed in these networks is key
• Standardize machine-readable presence: schema.org Restaurant is a simple foundation
• "Menu + constraints + ambiance" data is a conversion lever

For restaurants looking to optimize their AI visibility, our guide on agentic optimization details concrete steps.

For e-commerce merchants

The opportunity is massive: "connect once, be sellable everywhere".

• WooCommerce is preparing via MCP and agentic partnerships (Stripe). Our WooCommerce AI offering integrates these capabilities.
• The "orchestration layer" role between existing stacks and agentic channels is strategic
• Stripe has announced a suite aimed at reducing integration fragmentation

Action plan: four horizons

Ready to start this audit for your WordPress or WooCommerce site? Contact us for a free diagnostic.

Timeline: key milestones in agentic commerce

• January 2025 — OpenAI Operator (preview)
• May 2025 — PayPal × Perplexity (agentic commerce)
• September 2025 — Instant Checkout + ACP (Stripe/OpenAI)
• October 2025 — Apps in ChatGPT (SDK, built on MCP)
• December 2025 — Stripe Agentic Commerce Suite
• January 2026 — UCP (Universal Commerce Protocol, Google)
• March 2026 — Lemrock (France) raises €6M for agentic commerce

FAQ — Agentic commerce and marketplaces

What is agentic commerce? Agentic commerce refers to commercial transactions where AI doesn't just advise, but plans, calls tools/APIs, chains steps and can act (book, pay, modify, cancel) with variable user control.

Will marketplaces disappear with AI? No. Marketplaces don't mechanically disappear — they are restructured. Three roles emerge: disintermediation on simple segments, "platforms-as-API" behind third-party agents, or "platforms-as-agents" with their own interface. Value shifts toward execution, trust and data.

What is the MCP protocol and why does it matter for e-commerce? MCP (Model Context Protocol) is a connection standard between AI agents and external systems. OpenAI built its ChatGPT apps on it, and WooCommerce already exposes capabilities (products, orders) via an MCP endpoint. It's the foundation of agent-ready integration.

Does agentic commerce affect my restaurant? Yes. Google AI Mode deploys agentic booking capabilities via partners. To be visible and bookable, your restaurant needs to be correctly exposed (hours, menu, availability) with proper schema.org presence.

How does Busony help me become agent-ready? Busony offers an "agent-ready" audit and action plan covering structured data, MCP integration, GEO visibility and agentic instrumentation. Get in touch for a free diagnostic.