Agentic commerce · machine-readable commerce · AI standards · Author of Selling to Agents (Affero Press, 2026) · Milan
I work on how AI agents discover, read and buy from merchants, and on what merchants need to publish to be understood. Ex-founder: 1,600+ customers, 3,000 apps published on the app stores.
Agentic Standards Map: an open map of the standards work on how AI agents find, trust and transact with sellers, across W3C, the IETF, GS1, UN/CEFACT, the Agentic AI Foundation and the industry protocols. Every entry is sourced and dated, and published as structured data. frmoretto/agentic-standards-map · DOI 10.5281/zenodo.23357361
Layered Agentic Retrieval (LAR): a convention for how a business publishes what AI agents need to find, read and act on. Technically, a declarative publisher-side surface: it describes how an institution structures what it publishes so that agents can resolve, parse and act on it deterministically.
- Working paper: Marinoni Moretto, F. (2026). Layered Agentic Retrieval: A Publishing Architecture for Agent-Native Web Surfaces. SSRN. doi.org/10.2139/ssrn.6801118
- Presented at the W3C / GS1 Workshop E-commerce for Humans and AI Agents, Zurich, September 2026 (see Standards below)
- Reference implementation:
lar-spec/lar(JSON Schema and worked examples, Apache-2.0)
ucp-probe: a minimal, unaffiliated probe of agentic commerce endpoints, run against live stores since May 2026. MIT. Census of 19 September 2026: DOI 10.5281/zenodo.22893539
- Agentic AI Foundation, Agentic Commerce Working Group: presented the Agentic Standards Map to the working group, October 2026.
- W3C / GS1 Workshop, E-commerce for Humans and AI Agents, Zurich, September 2026. Position statement and slides: Toward a Shared Convention for the Publisher-Side Agent Surface.
- W3C: participant in the AI Agent Protocol and Web Machine Learning Community Groups; contributor to the draft Agentic Web Working Group charter (w3c/charter-drafts#888, merged).
- Universal Commerce Protocol: Discussion #867, Four measurements of the UCP surface, no shared denominator: a proposal for a common reporting frame.
- IETF DAWN: participant in the discussion on agent discovery and trust artifacts.
Selling to Agents: The Merchant Playbook for Agentic Commerce, Affero Press, 2026. Out now: buy it.
- AGNTCon + MCPCon North America, San Jose: Reading is Free, Spending is Not (Oct 23, 2026, upcoming)
- Agentic AI Foundation, Agentic Commerce Working Group: the Agentic Standards Map (Oct 2026)
- DINAMICA Workshop, LIUC Università Carlo Cattaneo, Castellanza: Perverse Incentives in Academic AI Detection: From Inferring to Declaring AI Use (Oct 2026)
- W3C / GS1 Workshop, Zurich: Toward a Shared Convention for the Publisher-Side Agent Surface (Sep 2026)
- ILTA Legal Tech On Tour, Milan: fireside on practical AI adoption in the legal profession (Jun 2026)
- Agents Anonymous Milan: Stream Coding methodology (Mar 2026)
- Claude Code Meetup Milan: Hardstop safety plugin (Feb 2026)
- Aperitivo AI Milan: Stream Coding methodology (Dec 2025)
- AI Tinkerers Milan: voice AI live demo (Jun 2025)
Hardstop: fail-closed safety plugin for Claude Code and Cowork. Regex patterns plus semantic analysis with Haiku.
npx hardstop install
Clarity Gate: pre-ingestion verification of epistemic quality for RAG.
Stream Coding: documentation-first AI development methodology, with specverify for independent specification verification and coderec for brownfield codebases.
Shopify Dev MCP Changelog: independent changelog for @shopify/dev-mcp, which ships almost weekly without release notes. Every version installed and diffed, updated daily by a GitHub Action.
ArXiParse: scientific paper extraction platform.
Claude Memory Research: systematic testing of Claude's memory system.
Claude's Hidden Toolkit: reverse-engineering of Claude.ai's internal tools, published under N1AI.
- Chapter Lead, The AI Collective Milan (2026-present)
- AI Practice & Community Lead, N1AI (2025-present)
sellingtoagents.com · Agentic Standards Map · SSRN · ORCID 0009-0004-0096-4712 · lar-spec · Sessionize · LinkedIn



