Why Now
Multiple independent signals are converging within 18 months.
A single signal is anecdotal. Three coordinated signals, agent protocols, payment rails, demand for semantic tooling, demand that the topic be addressed now, not two fiscal years from now.
Signal 1, Agent protocols have left the lab
Established Anthropic released MCP (Model Context Protocol) and put it into production. Google is pushing A2A. OpenAI exposes the Apps SDK in the ChatGPT client. These protocols transform the question "how does an agent talk to my system" from an open problem into an integration problem.
The strategic consequence is simple: a deterministic path now exists for a third-party agent to consume a merchant catalog. That path did not exist in 2023.
Signal 2, Payment rails are beginning to accept a delegated buyer
Emerging Stripe unveiled Agentic Commerce (ACP): typed intents, limits, agent identity, specific payment tokens. Visa announced Intelligent Commerce on the same principle for the card network side. Mastercard aligned with similar approaches.
This signal is decisive because it unlocks the transactional part. An agent without a compatible rail cannot close a deal; you plateau at "enriched recommendations." With these rails, a delegated purchase moves from prototype to production.
Signal 3, Pressure on catalog quality is rising
Established Google Merchant Center has tightened its requirements, Amazon is hardening listing rules, Perplexity and ChatGPT Search already index product pages with a bias toward enriched listings (schema.org, well-typed feeds).
Merchants who neglect semantic richness are not penalized brutally; they are silently demoted in synthesized answers. They stop being the answer.
Signal 4, Consumer assistants are crossing the pre-intent threshold
Emerging Commercial queries on ChatGPT, Perplexity, Copilot and Gemini are growing fast. The same assistants are executing increasingly transactional tasks (search, compare, reserve, compare support experiences). The moment a user fully delegates a recurring purchase to an agent is no longer unlikely.
Signal 5, The tooling market is taking shape
Prospective A "merchant readiness" layer is appearing: addressability scoring, feed audits, MCP exposure, policy SDKs. This layer is too young to be a consolidated category, but it explains why a shared reading framework, UCP, has immediate utility.
The timing argument, summarized
| Period | What is taking hold | What an operator should have done |
|---|---|---|
| 2024-2025 | MCP, ACP, Apps SDK arrive. SEO pressure + semantic feed enrichment. | Readiness audit, layer mapping, 2026 data budget allocation decision. |
| 2026 | First agentic use cases in production, starting with replenishment and support. Gen-1 agentic payment rails. | Semantic layer brought up to standard, typed policy governance, first MCP exposures. |
| 2027-2028 | Significant delegated purchases in selected verticals. Mature readiness tooling. | Agent intent observability, policy A/B testing, agent-specific KPIs at the executive level. |
Maturity legend: Established / Emerging / Prospective.
What this means for decision-making
Three decisions become actionable in 2026: (i) budget prioritization between SEO overhaul and semantic enrichment, (ii) technical opening of endpoints for agents, (iii) clarification of pricing and eligibility policies.
The logical next step is the audit methodology: a scoring framework for each of these decisions.