Definition
Agentic Merchant Protocol (AMP) is a merchant-side protocol and operating layer announced by Azoma on March 12, 2026. In Azoma’s current usage, AMP is designed to help brands and retailers keep control of product catalogue data, brand rules, compliance requirements, and product representation as AI agents discover, reason over, recommend, and potentially purchase products. Azoma positions AMP as sitting above protocols such as OpenAI’s Agentic Commerce Protocol (ACP) and Google’s Universal Commerce Protocol (UCP), rather than replacing them.
In marketing terms, AMP is about making product and brand information more usable by machines. As commerce shifts from webpages and search results toward conversational interfaces and AI shopping agents, structured product intelligence, governed messaging, and visibility into citations and rankings become part of the marketing stack rather than a purely technical concern. Microsoft describes conversation as the new “front door” to retail, and BCG notes that retailers risk losing direct engagement if third-party AI platforms become the main gatekeepers.
There is no standard formula for AMP itself. A practical way to measure AMP-related performance is through operational metrics such as:
Agentic share of voice = brand mentions in target agent outputs / total category mentions
Citation coverage = approved or canonical sources cited / total cited sources
Syndication coverage = eligible SKUs distributed to target agentic surfaces / total eligible SKUs
Agent-attributed conversion rate = orders from agentic channels / agentic visits or sessions.
These metrics align with the visibility, citation tracking, and attribution categories Azoma highlights in its agentic commerce materials.
How it relates to marketing
AMP matters because AI agents do not behave like human shoppers. They rely on structured product data, policies, availability, and external citations more than page layout, navigation, or headline copy. That shifts part of marketing from persuading humans on owned channels to shaping the machine-readable signals that influence agentic discovery, evaluation, and recommendation. OpenAI’s ACP documentation similarly emphasizes structured product feeds and checkout state as core inputs for agentic commerce flows.
For marketers, this means product content governance, merchant data quality, and AI-surface visibility become tightly connected. Azoma describes AMP as using canonical machine-native catalogues, open-web distribution, contextual prioritization, and analytics to influence how agents interpret products across multiple surfaces. That makes AMP relevant to ecommerce marketing, product information management, retail media strategy, SEO/GEO, and digital merchandising.
How to utilize it
A typical use of AMP would start with a canonical, machine-native product catalogue that includes product attributes, brand guidelines, compliance guardrails, persona signals, and availability rules. That information is then syndicated across agentic surfaces and the broader web so that AI agents encounter more consistent product intelligence wherever they reason. Azoma states that this is intended to reduce dependence on any single AI platform and improve consistency across agents and source domains.
Common use cases include high-SKU consumer goods, grocery, electronics, and other physical-product categories where accuracy, compliance, and representation matter. VentureBeat reports that Azoma is primarily focused on physical goods, especially CPG and FMCG, and not currently on areas such as SaaS, NFTs, or financial services.
In practice, a marketing or commerce team would use AMP to centralize product intelligence, distribute it to the channels and sources most likely to influence AI agents, monitor which products agents choose, review which outside sources are being cited, and then refine content, attributes, and distribution based on the findings. Azoma’s public materials describe visibility tracking, citation tracking, product ranking, and content syndication as key parts of that workflow.
Comparison to similar approaches
AMP belongs to the same emerging ecosystem as ACP, UCP, and AP2, but it addresses a different layer of the problem.
| Approach | Primary purpose | Main operating scope | Best understood as |
|---|---|---|---|
| AMP | Merchant-controlled catalogue intelligence, representation control, analytics, and syndication across agents and the open web. | Cross-agent and open-web merchant layer. | A merchant-side governance and distribution layer |
| ACP | Enables conversations between buyers, AI agents, and businesses to complete purchases; powers Instant Checkout in ChatGPT. | Open standard used for ChatGPT commerce and broader agent integrations. | A checkout and transaction protocol |
| UCP | Securely exchanges commerce data to support agentic ecommerce flows on Google surfaces such as AI Mode in Search and Gemini. | Google commerce ecosystem. | A Google-centered commerce orchestration protocol |
| AP2 | Secure, agent-led payments across platforms, including verifiable authorization and payment trust. | Payment layer across agentic commerce environments. | A payment and authorization protocol |
The simplest distinction is this: ACP and UCP help transactions happen inside specific ecosystems, AP2 helps payments happen securely, and AMP is positioned by Azoma as the merchant-side layer that governs what product intelligence agents should find, interpret, and use.
Best practices
Treat AMP as a governance and syndication layer, not as a substitute for basic feed hygiene, structured data, or checkout integrations. Merchants still need accurate product feeds, stable identifiers, current pricing and availability, and reliable commerce APIs. OpenAI’s commerce documentation is explicit that discoverability depends on structured feeds and accurate product data.
Keep brand, legal, and regulatory rules machine-readable wherever possible. Azoma’s public framing of AMP emphasizes compliance guardrails and enterprise-defined control, which suggests the value of AMP rises in categories where inaccurate claims, missing attributes, or inconsistent representation create risk.
Measure what agents actually do, not just what pages rank. That includes which products are surfaced, what sources are cited, how often approved content is used, and whether agentic traffic converts. BCG and Microsoft both point to the growing importance of AI-mediated discovery, which makes visibility and attribution in these environments more important than traditional last-click thinking alone.
Future trends
Public research suggests agentic commerce will become a significant channel rather than a side experiment. McKinsey estimates that US B2C retail alone could see up to $1 trillion in orchestrated revenue by 2030, and identifies protocols such as MCP, A2A, AP2, and ACP as part of the emerging infrastructure stack.
That points toward a likely future in which merchant-side layers like AMP coexist with checkout, payments, and agent-to-agent standards rather than replacing them. The more the commerce journey is mediated by AI systems, the more brands will need interoperable layers for product intelligence, consent, identity, payments, and performance measurement.
A second trend is that marketing teams will need to manage the “invisible shelf.” Product discoverability will depend less on the design of a single PDP and more on how consistent, complete, and trustworthy a product’s machine-readable footprint is across feeds, cited sources, and agentic channels. In that environment, protocols and operating layers like AMP are likely to be evaluated on whether they improve representation control, reduce ambiguity, and produce measurable uplift in agentic visibility and conversion.
Related Terms
Agentic commerce
Agentic Commerce Protocol (ACP)
Universal Commerce Protocol (UCP)
Agent Payments Protocol (AP2)
Agentic Commerce Optimization (ACO)
Generative Engine Optimization (GEO)
Answer Engine Optimization (AEO)
Product information management (PIM)
Structured product feed
Digital merchandising