UCP Ancillaries Extension Proposal: Add-Ons for Agentic Commerce

A new proposal brings structured support for optional add-ons, upgrades, and cross-sells to the Universal Commerce Protocol — essential infrastructure for travel, hospitality, and beyond.

About the Contributor

Tobias Alldén (alldentobias) returns to UCP after his previous ancillaries proposal in February. His continued focus on add-on commerce suggests deep domain expertise, likely from travel or hospitality technology. The persistence signals genuine industry need for standardized ancillary handling in agentic transactions.

Why Ancillaries Matter

When an AI agent books a flight or hotel for a user, the base transaction is just the beginning. Airlines make significant revenue from seat upgrades, baggage, and meal selections. Hotels offer room upgrades, spa packages, and late checkout. E-commerce has warranty extensions, gift wrapping, and expedited shipping.

Currently, there's no standardized way for merchants to present these options or for AI agents to select them on behalf of users. Each vertical handles add-ons differently, forcing agents to implement custom logic for every merchant.

What the Extension Proposes

The ancillaries extension introduces a structured schema for optional add-ons that can attach to any UCP transaction:

Ancillary Types: Categorization (upgrade, add-on, protection, service) enabling agents to understand what they're selecting
Availability Rules: When ancillaries can be selected (at booking, pre-arrival, during stay) and what they attach to
Pricing Models: Per-item, per-person, per-night, and complex dynamic pricing support
Dependencies: Some ancillaries require others (e.g., premium seat requires lounge access for full experience)
Mutual Exclusivity: Some ancillaries conflict (e.g., standard vs. premium room upgrade)

Technical Implementation

The proposal extends the checkout flow with an optional ancillaries array:

{
  "checkout": {
    "items": [...],
    "ancillaries": {
      "available": [
        {
          "id": "seat-upgrade-1a",
          "type": "upgrade",
          "name": "Premium Seat 1A",
          "description": "Extra legroom, priority boarding",
          "pricing": {
            "amount": "75.00",
            "currency": "USD",
            "model": "per_segment"
          },
          "attachesTo": "line_item_0",
          "availableUntil": "2026-03-15T10:00:00Z"
        }
      ],
      "selected": ["seat-upgrade-1a"]
    }
  }
}

Why This Matters for AI Agents

Structured ancillaries enable several agent capabilities that are currently impossible or require extensive custom integration:

Preference Learning: "Patrick always selects premium economy when available under $100" becomes learnable when ancillaries are standardized
Budget Optimization: Agents can evaluate all available upgrades against remaining budget
Bundle Detection: Identifying when selecting multiple ancillaries triggers package pricing
Cross-Merchant Comparison: "Flight A is cheaper but Flight B includes baggage" becomes computable

Relationship to Previous Work

This proposal builds on Alldén's February contribution and the Hotels extension from Sabre. Together, they suggest UCP's vertical extension model is gaining traction:

Hotels Extension (Feb 16): Room types, amenities, policies
Ancillaries (Feb 17, expanded Mar 2): Add-ons applicable across verticals
Pattern Emerging: Domain experts contributing vertical-specific schemas while keeping core protocol stable

Next Steps

The proposal is currently open for review. Key questions for the working group:

Should ancillary selection happen during checkout or require separate operations?
How should cancellation/refund policies for ancillaries be represented?
What's the interaction model with payment instruments that have ancillary restrictions?

For AI agents handling real purchasing decisions, ancillaries represent a significant revenue opportunity for merchants and a genuine UX improvement for users. Getting the schema right matters.