Sunday, June 14, 2026 Next update in 6d 20h 14m 37s

— A weekly publication —

The Agentic Commerce Report

A weekly read of everything that moved in agentic commerce — protocols, payment rails, retailer pilots, regulation. Summarised, sourced, and stitched to what came before.

Standards & Protocols

15 events tracked

Open interoperability specifications for agentic commerce are advancing in parallel across foundation, vendor, and network venues. Tracked weekly here: every spec publication, working-group formation, and interop-event outcome.

launch

Stripe Projects adds new agent integrations, more providers, and custom developer controls

Stripe Projects added 16 providers (49 total) and agent integrations with Hermes, Factory Droids, and Warp on June 11.

Stripe Projects enables AI agents to provision cloud infrastructure (hosting, databases, authentication, billing) from the command line without human dashboard interaction. Agent traffic reached 40% of Stripe's documentation requests in 2025, and 70% of CLI calls for API resources now come from agents. The June 11 update added three integrations: Hermes (Nous Research's open-source persistent agent), Factory Droids, and Warp — all model-agnostic coding agents. Per-provider spending caps and named environment isolation (development, staging, production) provide the first enterprise-grade guardrails for agent-provisioned software published by a major payments infrastructure company. Cloudflare Radar reported agent traffic surpassed human internet traffic for the first time the week of this announcement. Stripe's Agentic Commerce Suite (Sessions 2026) and Projects together span the full agent workflow: infrastructure provisioning, checkout, fraud detection (Radar bot-score, 2026-w22), and authorization optimization (DCAP, 2026-w23).

  1. Stripe Blog
launch

Claude Fable 5 and Claude Mythos 5

Anthropic launched Claude Fable 5 for general use and Mythos 5 (restricted) at $10/$50 per million tokens on June 9.

Fable 5 is Anthropic's first Mythos-class model made generally available, replacing Claude Mythos Preview (restricted to Glasswing partners since April 2026; 2026-w23). Mythos 5 is the same underlying model with cybersecurity safeguards lifted; it is restricted to Glasswing partners and select biology researchers. Fable 5 ranks highest among frontier models on Cognition's FrontierCode benchmark and achieves 84% on Online-Mind2Web (autonomous web navigation including checkout flows). Pricing at $10 per million input and $50 per million output tokens is less than half the cost of Mythos Preview. Anthropic required 30-day data retention for all Mythos-class model traffic. Stripe reported Fable 5 completed a 50-million-line Ruby codebase migration in one day. Three days after launch, the US government issued a directive suspending both models (2026-w24-regulation-us-directive-fable-mythos).

  1. Anthropic News
M&A

Anthropic confidentially submits draft S-1 to the SEC

Anthropic submitted a confidential draft S-1 registration statement to the SEC for a proposed IPO.

The S-1 filing follows Anthropic's $65B Series H (2026-w22), which valued the company at $965B post-money and disclosed $47B in annualised revenue. A confidential S-1 submission under the JOBS Act starts the SEC review clock without committing to a public offering timeline. Anthropic is the third major AI company to file an S-1 in the 2026 window: SpaceX submitted a public S-1 on May 20 targeting a June Nasdaq listing, and OpenAI submitted its own confidential draft around May 22. No offering price or share count was set at the June 1 filing date. Anthropic builds and maintains the Model Context Protocol open standard and the Stainless SDK generation platform acquired in May (2026-w21), which generates official client libraries for hundreds of API providers.

  1. Anthropic News
M&A

Anthropic raises $65B in Series H funding at $965B post-money valuation

Anthropic closed a $65B Series H on May 28 at a $965B post-money valuation, disclosing $47B in run-rate revenue for the first time.

The round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, with $15B from previously committed hyperscaler capital including Amazon. The $47B ARR figure is the first public revenue milestone Anthropic has published and the largest reported by any AI-native software company. Anthropic builds and maintains Model Context Protocol (MCP), the open standard used by the majority of commercial agent integrations. The Stainless acquisition (2026-w21) brought SDK and MCP server generation in-house the week before this raise. The Series H also includes strategic infrastructure positions from Micron, Samsung, and SK hynix. The round extends Anthropic's position as the publisher of MCP at a moment when three distinct agent protocol stacks — MCP, A2A, and UCP — are competing for enterprise adoption across agentic commerce deployments.

  1. Anthropic News
launch

Introducing Claude Opus 4.8

Anthropic released Claude Opus 4.8 on May 28, scoring 84% on Online-Mind2Web and completing every case in the Super-Agent benchmark, beating GPT-5.5 at parity cost.

Online-Mind2Web measures autonomous web navigation including checkout flows; the 84% score is described as a meaningful jump over Opus 4.7 and sets the new high for any deployed browser-agent model. Claude Opus 4.8 is also the first model to complete every end-to-end case in the Super-Agent benchmark. Dynamic Workflows, a research-preview feature in Claude Code, runs hundreds of parallel subagents within a single session — enabling automation at the scale required for multi-merchant checkout orchestration and catalog-wide AEO operations. Pricing is unchanged from Opus 4.7 at $5 per million input tokens; fast mode is now three times cheaper than it was for prior models. The Stainless acquisition (2026-w21) means Anthropic now controls the SDK tooling used to distribute MCP server connections; Opus 4.8's browser-agent improvements compound with that distribution layer for agents operating in checkout or discovery contexts.

  1. Anthropic News
spec

How we're helping retailers thrive with new Universal Commerce Protocol features and AI tools on Google

Google added multi-item carts, Identity Linking, and BNPL to UCP; expanded internationally to Canada, Australia, and UK.

The Google Marketing Live additions push the Universal Commerce Protocol (2026-w02-standards-google-ucp-launch) past single-item checkout into multi-item carts, Identity Linking, and BNPL, completing the feature parity gap with platform-native checkouts on Amazon and Shopify. The Canada, Australia, and UK expansion is UCP's first geographic move outside the US launch geography and parallels Mastercard Agent Pay's Australian transaction (2026-w05-pilots-mastercard-australia-first-agentic-payment) and Santander's European first (2026-w10-pilots-santander-mastercard-europe-first-payment). Identity Linking ties to the FIDO Agentic Auth working group (2026-w18-standards-fido-agentic-working-groups) and OpenID4VP self-certification (2026-w09-identity-openid-vc-self-certification). The release accompanies the Universal Cart consumer launch (2026-w21-aeo-google-universal-cart) and the expanded UCP Tech Council (2026-w17-standards-ucp-tech-council-expansion). GML is Google's annual advertiser event, so the release links UCP feature additions to the advertising-and-monetisation stack rather than the engineering-conference cadence used for the original UCP launch.

  1. Google Blog
M&A

Anthropic acquires Stainless

Anthropic acquired Stainless, which generates SDKs and MCP servers for OpenAI, Google, and hundreds of other companies.

The acquisition is Anthropic's first developer-tooling M&A in this archive and consolidates the SDK-and-MCP-server generation layer that sits beneath every major API vendor — Stainless customers include OpenAI, Google, and hundreds of other companies. MCP (2024-w48-standards-anthropic-mcp, 2025-w48-standards-mcp-spec-update) remains the substrate; Stainless industrialises its production for the broader ecosystem. The deal sits alongside MCP's flagship-project role inside the Linux Foundation AAIF (2025-w50-standards-linux-foundation-aaif) — Anthropic owns the spec stewardship via the foundation while operating the leading SDK generator commercially. The acquisition extends the Standards lane's M&A footprint, which previously consisted only of protocol launches and governance formations such as the UCP Tech Council expansion (2026-w17-standards-ucp-tech-council-expansion).

  1. Anthropic News
spec

FIDO Alliance to Develop Standards for Trusted AI Agent Interactions

FIDO Alliance formed Agentic Auth TWG backed by Google AP2 and Mastercard Verifiable Intent; 60+ orgs committed.

The Agentic Auth Technical Working Group is the first FIDO-led standards body scoped specifically to AI agents, extending FIDO2 and passkey infrastructure into agent authentication. Google AP2 (2025-w38-standards-google-ap2-protocol) and Mastercard Verifiable Intent contribute the protocol-level inputs, paralleling the OpenID Foundation's AI identity whitepaper (2025-w41-identity-openid-foundation-ai-whitepaper) and the OpenID4VP self-certification programme (2026-w09-identity-openid-vc-self-certification). The 60+ committed organisations span identity vendors, browser makers, and payment networks, structurally echoing the Linux Foundation AAIF formation (2025-w50-standards-linux-foundation-aaif) and the UCP Tech Council expansion (2026-w17-standards-ucp-tech-council-expansion). FIDO becomes the third multi-stakeholder governance body for agentic infrastructure formed in this archive, occupying the authentication layer. The working group's outputs feed into the Identity Linking capability Google added to UCP at I/O 2026 (2026-w21-standards-google-ucp-gml).

  1. FIDO Alliance
launch

Amazon, Meta, Microsoft, Salesforce, and Stripe join Universal Commerce Protocol Tech Council

Ten-member governance body now spans retail, social commerce, cloud infrastructure, enterprise software, and payments, signalling broad industry convergence on UCP.

The five additions extend the Universal Commerce Protocol Tech Council (2026-w02-standards-google-ucp-launch) from a retail-and-payments group into a cross-category governance body. Amazon's inclusion is structurally notable given its first-party agent posture in Rufus (2025-w47-pilots-amazon-rufus-auto-buy) and its litigation against Perplexity's Comet browser (2026-w11-regulation-amazon-perplexity-court-order). Meta brings social commerce, Microsoft brings the Copilot Checkout consumer surface (2026-w02-payments-microsoft-copilot-checkout), Salesforce brings enterprise CRM, and Stripe brings the processor layer it already operates across ACP (2025-w40-payments-stripe-openai-acp-instant-checkout) and its own Suite (2025-w50-payments-stripe-agentic-commerce-suite). The expansion places UCP alongside the Linux Foundation's AAIF (2025-w50-standards-linux-foundation-aaif) and the FIDO Agentic Auth working group (2026-w18-standards-fido-agentic-working-groups) as a multi-stakeholder governance body. The council now totals ten members.

  1. Newsfile Corp
spec

Google launches Universal Commerce Protocol at National Retail Federation

Open-source standard covering the full shopping journey co-launched with Shopify, Walmart, Target, and Etsy, endorsed by 20+ partners including Mastercard, Visa, and Stripe.

The Universal Commerce Protocol (UCP) is the first open standard scoped to the full shopping journey rather than a single leg, layering on top of A2A (2025-w15-standards-google-a2a) for inter-agent transport and AP2 (2025-w38-standards-google-ap2-protocol) for payments. The Shopify, Walmart, Target, and Etsy co-launch covers four of the largest US merchant ecosystems; Mastercard, Visa, and Stripe inclusion places UCP across all major payment rails. UCP later expands with multi-item carts, Identity Linking, and BNPL (2026-w21-standards-google-ucp-gml) and brings on Amazon, Meta, Microsoft, Salesforce, and Stripe to its Tech Council (2026-w17-standards-ucp-tech-council-expansion). The Universal Cart launch at I/O 2026 (2026-w21-aeo-google-universal-cart) is the first consumer surface built on UCP, and Ulta Beauty (2026-w17-pilots-ulta-beauty-google-gemini-checkout) is the first retailer to ship UCP checkout.

  1. TechCrunch
launch

Linux Foundation forms Agentic AI Foundation with MCP, goose, and AGENTS.md

AAIF launched with Anthropic's MCP, Block's goose, and OpenAI's AGENTS.md as flagship projects; AWS, Google, and Microsoft among platinum members.

The Agentic AI Foundation (AAIF) places the Model Context Protocol (MCP) (2024-w48-standards-anthropic-mcp), goose, and AGENTS.md under a neutral, vendor-independent governance body — the same structural move the Linux Foundation made for Kubernetes, OCI, and OpenJS. The platinum-member roster covers every major cloud provider, removing single-vendor capture risk from the agent stack's lower layers. The formation parallels the Universal Commerce Protocol Tech Council (2026-w02-standards-google-ucp-launch, 2026-w17-standards-ucp-tech-council-expansion) and the FIDO Agentic Auth working group (2026-w18-standards-fido-agentic-working-groups), creating three multi-stakeholder governance bodies covering different layers: AAIF for tool-and-discovery, UCP for commerce flows, FIDO for authentication. MCP's anniversary spec (2025-w48-standards-mcp-spec-update) shipped two weeks earlier as part of the transition to neutral stewardship. Block's goose contribution adds a developer-facing CLI agent alongside the protocol-level MCP.

  1. Linux Foundation
spec

Model Context Protocol publishes 2025-11-25 specification

One-year anniversary spec adds async Tasks, improved OAuth support, and extensions; first revision with backward-compatibility guarantees.

The 2025-11-25 revision is the first Model Context Protocol (MCP) spec update to ship a backward-compatibility guarantee, marking the protocol's transition from rapid iteration to stable infrastructure one year after its original release (2024-w48-standards-anthropic-mcp). Async Tasks address the long-running-operation gap that earlier MCP servers worked around with polling. OAuth refinements align MCP authentication with the OpenID Foundation's identity work (2025-w41-identity-openid-foundation-ai-whitepaper). The release precedes MCP's adoption as a flagship project of the Linux Foundation's Agentic AI Foundation (2025-w50-standards-linux-foundation-aaif) by two weeks, and the spec is the transport that the Stripe-OpenAI Agentic Commerce Protocol (2025-w40-payments-stripe-openai-acp-instant-checkout) and Google's AP2 (2025-w38-standards-google-ap2-protocol) both compose on top of. The anniversary timing makes the release an explicit governance checkpoint for the protocol.

  1. Model Context Protocol Blog
spec

Google launches Agent Payments Protocol (AP2) with 60+ partners

Open payment protocol using cryptographically-signed Mandates to authorise agent-led transactions across credit, debit, stablecoins, and real-time transfers; extends A2A and MCP.

AP2 extends the A2A inter-agent layer (2025-w15-standards-google-a2a) and the MCP tool layer (2024-w48-standards-anthropic-mcp) into payment authorisation, using cryptographically-signed Mandates that record human intent, agent identity, and transaction scope. The 60+ partner roster spans card networks (Mastercard, American Express), processors (Adyen, Worldpay), stablecoin issuers, and real-time payment operators, making AP2 the first agent payment spec with cross-rails coverage. The Mandate model later anchors the FIDO Alliance Agentic Auth working group (2026-w18-standards-fido-agentic-working-groups). AP2 sits beneath the Universal Commerce Protocol (2026-w02-standards-google-ucp-launch) as the payment leg of the full commerce stack and informs Visa's parallel Trusted Agent Protocol (2025-w42-payments-visa-trusted-agent-protocol) released a month later.

  1. Google Cloud Blog
spec

Google publishes Agent2Agent protocol at Cloud Next

Open protocol for inter-agent communication and capability negotiation, launched at Google Cloud Next with support from 50+ technology and services partners.

The agent-to-agent (A2A) protocol occupies the inter-agent layer that MCP (2024-w48-standards-anthropic-mcp) does not address: MCP standardises how an agent reaches tools and data, while the A2A protocol standardises how two agents discover, authenticate, and negotiate capabilities with each other. The 50+ launch partners include Salesforce, SAP, ServiceNow, and Atlassian, placing enterprise workflow vendors alongside cloud providers in the spec's governance. A2A becomes the transport that Google's Agent Payments Protocol (2025-w38-standards-google-ap2-protocol) extends with payment Mandates, and the Universal Commerce Protocol (2026-w02-standards-google-ucp-launch) layers full commerce flows on top. The protocol is later folded into the Linux Foundation Agentic AI Foundation governance discussion (2025-w50-standards-linux-foundation-aaif), joining MCP and AGENTS.md under neutral stewardship.

  1. Google Developers Blog
spec

Anthropic releases Model Context Protocol

Open spec for connecting AI assistants to data sources and tools, published with reference servers in TypeScript and Python.

The Model Context Protocol (MCP) is the first cross-vendor specification for the tool-access layer of agent infrastructure, predating Google's agent-to-agent (A2A) protocol (2025-w15-standards-google-a2a) by roughly five months and the Agent Payments Protocol (2025-w38-standards-google-ap2-protocol) by ten. The reference servers ship under the MIT licence with bindings in TypeScript and Python, providing the substrate that later commerce specs layer on top of — AP2 cites MCP as a transport, and the Stripe-OpenAI Agentic Commerce Protocol (2025-w40-payments-stripe-openai-acp-instant-checkout) uses MCP-shaped tool calls. The 2025-11-25 anniversary revision (2025-w48-standards-mcp-spec-update) adds async Tasks and OAuth refinements. MCP is also one of three flagship projects under the Linux Foundation's Agentic AI Foundation (2025-w50-standards-linux-foundation-aaif), establishing neutral governance for the layer.

  1. Anthropic Engineering

What standards govern agentic commerce?

Agentic-commerce standards are the open specifications that let agents from different vendors interoperate — discovering tools, exchanging payment authorisations, and proving identity across a transaction. The active set includes Anthropic's Model Context Protocol (MCP) for tool exposure, Google's Agent-to-Agent (A2A) and Agent Payments Protocol (AP2) for inter-agent messaging and payment intent, the Universal Commerce Protocol (UCP) led by Google and Shopify for end-to-end agent checkout, and OpenID Foundation specifications (OpenID4VP, OpenID4VCI, HAIP 1.0) for verifiable credentials. These are draft or v0.x at most major networks; convergence across vendors is the principal open question.

Open interoperability specifications for agentic commerce are advancing in parallel across foundation, vendor, and network venues. Anthropic published the Model Context Protocol in November 2024; the one-year-anniversary spec in November 2025 added asynchronous Tasks and improved OAuth handling. Google published the Agent-to-Agent protocol at Cloud Next in April 2025 and followed in September 2025 with the Agent Payments Protocol backed by 60-plus partners. The Linux Foundation formed the Agentic AI Foundation in December 2025 with MCP, Block's goose, and OpenAI's AGENTS.md as founding projects. Google and Shopify co-launched the Universal Commerce Protocol at NRF in January 2026; Amazon, Meta, Microsoft, Salesforce, and Stripe joined its ten-member Tech Council in April 2026. The FIDO Alliance formed agentic-commerce working groups in April 2026, accepting AP2 v0.2 and Mastercard's Verifiable Intent as founding contributions. No specification has reached v1.0 across a multi-vendor implementer base; most live deployments run on draft or vendor-specific variants. This hub tracks every spec publication, working-group formation, and interop-event outcome across the open-standard layer.

What is the Model Context Protocol (MCP)?

Anthropic published the Model Context Protocol in November 2024 as an open specification for connecting AI assistants to external data sources and tools over HTTP. The protocol defines how an AI model discovers available tools, invokes them, and receives structured results — covering the complete tool-exposure layer from capability announcement to error handling. Any server implementing the protocol can expose its capabilities to any MCP-compliant model, regardless of vendor.

The one-year anniversary specification, published in November 2025, added three capabilities not present in the initial release: asynchronous Tasks for long-running operations, improved OAuth handling for authenticated tool access, and a standardised sampling interface. The Linux Foundation's Agentic AI Foundation, formed in December 2025, adopted MCP as a founding project alongside Block's goose and OpenAI's AGENTS.md. No version of the specification has reached v1.0 across a multi-vendor implementer base as of April 2026.

Sources: Anthropic — Model Context Protocol announcement (Nov 2024)

What is the A2A protocol?

Google published the Agent-to-Agent protocol at Google Cloud Next in April 2025 as an open specification for inter-agent communication. The protocol defines how one AI agent discovers another agent's capabilities, sends task requests, and receives structured responses — covering the message-layer between autonomous agents regardless of which model or platform each agent runs on.

In September 2025, Google followed with the Agent Payments Protocol (AP2), a companion specification defining how an agent communicates payment intent within an A2A message flow. AP2 launched with backing from 60-plus partners including Mastercard, Visa, and Stripe. The FIDO Alliance formed agentic-commerce working groups in April 2026, accepting AP2 v0.2 and Mastercard's Verifiable Intent framework as founding contributions. The A2A protocol and AP2 are both at draft stage; no joint v1.0 release has been published across a multi-vendor implementer base as of April 2026.

Sources: Google — A2A protocol announcement (Apr 2025)

What is the Universal Commerce Protocol (UCP)?

Google and Shopify co-launched the Universal Commerce Protocol at the NRF Retail Conference in January 2026 as an open specification for end-to-end AI-agent checkout. The protocol defines how an agent discovers a merchant's catalogue, selects a product, and completes an authenticated payment transaction in a single structured flow — spanning product data, pricing, payment authorisation, and order confirmation.

In April 2026, Amazon, Meta, Microsoft, Salesforce, and Stripe joined the UCP Tech Council, bringing its membership to ten founding organisations. The April announcement expanded the protocol's scope to include agent-to-agent cart handoff, where one agent passes a filled cart to another agent operating at a different merchant. The specification remains in draft form; Google and Shopify published no timeline for a stable v1.0 release as of April 2026.

Sources: Google — Universal Commerce Protocol at NRF (Jan 2026)

What is the Agentic AI Foundation?

The Linux Foundation formed the Agentic AI Foundation in December 2025 to provide neutral governance for open agentic-AI specifications. The Foundation launched with three founding projects: Anthropic's Model Context Protocol (MCP), Block's goose (an open-source agentic CLI), and OpenAI's AGENTS.md (a specification format for defining agent capabilities in a machine-readable file). Each project retained its existing governance structure under the Foundation's IP and contribution policies.

The Foundation's mandate covers specification development, conformance testing, and developer tooling for the agentic AI layer — the protocols and runtimes that govern how autonomous agents discover tools, communicate with each other, and execute multi-step tasks. The Linux Foundation hosts over 900 open-source projects and provides the legal, marketing, and technical governance infrastructure. No first joint specification under Agentic AI Foundation governance had reached a stable release as of April 2026.

Sources: Linux Foundation — Agentic AI Foundation (Dec 2025)