The Agentic Commerce Frontier 📅 | June 9 - June 22 2026
Thank you for reading this edition of The Agentic Commerce Frontier. This is a longer newsletter than usual as happens every time the week after I publish an updated Agentic Commerce Ecosystem Map.
Thanks for the great conversations in the past couple of weeks with the CFO that is working on payment decision control, the BNPL provider who is building an Agentic BNPL aggregator, the German software provider building AI agents for retail and for customer service and the French banker investigating Agentic Subscription Management! Looking froward to more conversations!
🔥 TL;DR
Visa introduced agent scoring, an agent registry/directory, token improvements, stablecoin settlement enhancements, and an OpenAI collaboration aimed at secure payments inside agentic commerce experiences
My takeaway: Visa is positioning itself as both transaction rail and agent-trust infrastructure, not merely the credential behind checkout
Mastercard launched Agent Pay for Machines, extending its Agent Pay architecture to permissioned, machine-speed transactions across cards, real-time payments, account-to-account rails, and stablecoins
My takeaway: It’s becoming obvious that autonomous B2B and device-to-device commerce will scale before fully delegated consumer shopping matures
Adyen introduced Adyen Agentic, a modular API stack for agentic shopping across feed, cart, and payment functions, with early alignment from American Express, Mastercard, Salesforce, Visa, and several retailers
My takeaway: If you are to believe that agentic commerce will fragment before it standardizes, then Adyen is making the right bet
Shopify opened self-serve access to the Universal Cart Protocol and Catalog API, while enabling eligible merchants to make structured catalog data available to AI-powered shopping surfaces
My takeaway: The next commerce moat may be boring operational hygiene: clean inventory, structured policies, accurate variants, and machine-legible checkout rules
Checkout.com found fast-forming consumer demand but weak trust: 27% of consumers surveyed said they trust no organization to operate an AI shopping agent, while merchants reported preparing faster than current transaction volume suggests
My takeaway: Autonomy is not the benchmark … for now
🤖 Agentic Commerce Primer: Amazon v. Perplexity (Or the Right To Bring Your Own Agent)
Amazon v. Perplexity is not a shopping-bot dispute. It is the first serious test of whether consumers can delegate commercial intent — or whether platforms get veto power over the agents that act for them.
The core question in agentic commerce is simple: can a platform refuse your representative?
That question is now in court.
Amazon says Perplexity’s Comet agent accessed accounts covertly, masked automation as human browsing, and created security risks. Perplexity says Amazon is trying to stop users from choosing their own assistant. Both are partly right. Agents should not be able to enter commercial systems by pretending not to be agents. But if a customer authorizes software to act on their behalf, does the platform have to recognize that delegation? Or can it force every user back through its own interface, ranking logic, and checkout path?
That is the real fight.
Not AI versus ecommerce.
Not innovation versus incumbency.
Not bots versus websites.
This is about the right to bring your own agent.
The banned assistant
Imagine walking into a store with a personal assistant.
They know your size, budget, allergies, return habits, and loyalty accounts. You say: “Pick the best option and buy it for me.” The store says: you may enter, but your assistant may not.
Online, that is no longer hypothetical.
The assistant is an AI agent. The store is Amazon. And the question is whether that agent is acting as your representative or as a trespasser.
The case, briefly
Amazon sued Perplexity over Comet, its agentic browser, alleging covert account access, disguised automation, security risk, and ignored demands to stop. Reuters reported that Amazon’s complaint said the agent could log into shopping accounts and place orders on users’ behalf, while Perplexity countered that Amazon was trying to prevent users from choosing their own AI assistants. A federal judge temporarily blocked the tool in March 2026, and the Ninth Circuit is now examining who the relevant actor is, the user or the agent? and how a 1986 computer-fraud statute should apply to autonomous software. The legal question is narrow but the market impact is not.
Killing the lazy narratives
The first lazy take is that Amazon is anti-innovation. It is not. Amazon’s strongest argument is straightforward: agents that disguise themselves, bypass safeguards, or create unclear liability should not be allowed to operate inside a commercial system. Transparency is not protectionism. It is the minimum price of admission.
The second lazy take is that user consent makes everything permissible. It does not. Logged-in commerce is not the open web. It involves payment credentials, purchase history, addresses, subscriptions, returns, fraud exposure, and customer-service obligations. A user can delegate, but that delegation does not erase every duty owed to the platform, the merchant, or the broader financial system.
User consent is necessary.
It is not sufficient.
Platform sovereignty vs user agency
Amazon is not just a website. It is an economic machine.
It controls ranking, sponsored placement, comparison logic, bundling, Prime messaging, substitutions, review visibility, pricing context, returns, and post-purchase service. It shapes how intent is interpreted before a transaction happens. A user-appointed agent threatens that system because it changes who interprets intent.
Amazon is not only defending its infrastructure from a bot. It is defending the right to be the sole interpreter of customer intent inside its marketplace.
That is the deeper issue.
The fight is not just about access. It is about who owns the moment between desire and transaction.
The portable mandate
I’ve spoken many times about portability in agentic commerce: memory portability, profile portability, consent portability. Here comes the mandate portability.
A portable mandate is the user’s ability to authorize a software agent to act across commercial environments under clear, enforceable limits. “Compare prices, but do not buy.” “Buy household staples under 75 euros.” “Reorder only from merchants I have used before.” “Purchase only if delivery arrives before Friday.” “Use my card, but not my gift balance.” “Ask before substituting brands.”
This is what agentic commerce actually unlocks: not automation, but programmable intent.
The wrong defendant, the right principle
Perplexity is not a perfect standard-bearer.
If Amazon’s allegations are directionally correct — that the agent masked its identity or continued operating after access was revoked — courts are unlikely to reward that behavior. A right to delegate cannot be built on evasion.
But bad facts should not create a bad rule.
Perplexity may lose on conduct and still be directionally right on principle. Amazon may be right about deception and still wrong about platform veto power.
Recognized agents
The answer is not open access or total restriction. It is recognition.
A legitimate commercial agent should meet five conditions: identity, so the merchant can tell it is interacting with an agent; mandate, so the user’s authorization is provable and revocable; observability, so disputes and fraud can be investigated; accountability, so a known entity stands behind the system; and policy negotiation, so merchants can set rules without turning them into a disguised blanket ban on delegation.
Recognized agents are the middle ground between bot anarchy and platform feudalism.
Why this extends far beyond Amazon
If Amazon can broadly block user-appointed agents, every major platform can follow: airlines, banks, delivery apps, ticketing systems, insurers, travel marketplaces. Agentic commerce becomes a network of permissioned integrations controlled by incumbents.
If agents can access platforms simply because users share credentials, the opposite failure mode appears: opaque automation, degraded service, fraud, and operational chaos.
Neither model scales.
What the market needs is a trust layer for delegation.
Agents are not ready
There is another reality check here.
Current agents are still fragile. EComAgentBench, a June 2026 benchmark covering 662 shopping tasks, found that even top-performing models reached only 57.1 percent accuracy, with performance falling as user intent became more implicit or distributed across interaction steps. The point is not that agents are useless. The point is that the right to delegate cannot mean the right to be unprotected from your own agent.
Auditability, constraints, and revocation are not technical extras.
They are consumer protection.
The real economic stakes
This is not just about checkout. It is about retail media.
Agents do not experience interfaces the way humans do. They do not respond to banner placement, urgency cues, sponsored carousels, or carefully designed upsells. They compress the journey. They ask: what satisfies the mandate, what is cheapest, what arrives on time, what is reliable, and what should be ignored?
That is good for users.
It is disruptive for any business model built on shaping behavior before purchase.
Amazon’s conflict with Perplexity is not only about whether an agent can transact. It is about whether the platform can force intent through a monetized interface.
A limited right to delegate
The policy answer is not absolute openness.
It is a limited right.
Platforms should be able to block agents that are deceptive, insecure, abusive, or noncompliant with reasonable technical standards. But they should not be allowed to categorically block transparent, user-authorized agents simply because those agents reduce ad exposure or weaken interface control.
A platform should be able to reject a trespasser.
It should not be able to reject a customer’s representative merely because the representative is software.
The objections
“Amazon owns the platform. It can set the rules.”
Yes, but ownership becomes more complicated when the platform is also a gatekeeper to essential commercial activity. The question is not whether Amazon has rights. It is whether those rights include nullifying user agency as a category.
“Agents will create fraud and support nightmares.”
They will. That is why identity, logging, and accountability are non-negotiable. The answer to risk is standards, not prohibition.
“Users can already shop manually.”
That is not the point. The issue is not access to goods. It is the right to choose the intermediary that interprets your intent.
“This is just scraping in nicer clothes.”
Sometimes it is. That is why the distinction between recognized agents and covert automation matters. The principle worth defending is delegation, not evasion.
What the system could look like
A plausible near-future flow is already visible:
A user connects an agent to a wallet or identity provider
The agent receives a signed, scoped mandate
The merchant sees the agent’s identity and authorization
The merchant returns structured product, pricing, delivery, and policy data
The agent recommends or executes within its mandate
The transaction record includes agent identity, authorization state, and a verifiable log
The user can revoke access instantly.
The winners in this system may not be the best agents.
They may be the first networks that make mandates portable and trustworthy.
Closing
A century of commerce law assumes that people can act through representatives. The internet largely suppressed that idea by forcing users through controlled interfaces.
AI agents bring representation back, but in software form.
Amazon v. Perplexity will be argued as a computer-access case. It should be understood as something larger: the first serious test of whether commercial intent is portable. If the answer is no, agentic commerce becomes another platform-controlled channel. If the answer is yes without constraints, it becomes chaos.
The more interesting answer sits in between.
Consumers should be able to bring their own agents, but only agents willing to identify themselves, prove their mandate, leave an audit trail, and accept responsibility.
The future of shopping will not be decided by whether bots can click buttons.
It will be decided by whether platforms must recognize the customer’s chosen representative.
🚀 Major Announcements & Funding News
Visa expands agentic commerce infrastructure at Payments Forum: Visa announced Agent Scoring, Agentic Registry, Agentic Directory, a Large Transaction Model, token enhancements, stablecoin settlement updates, and a strategic OpenAI collaboration for secure payments within AI-powered commerce flows (Visa)
Mastercard launches Agent Pay for Machines: Mastercard introduced Agent Pay for Machines with early supporters including Adyen, Ant International, BVNK, Checkout.com, Cloudflare, Coinbase, Global Payments, OKX, Stripe, and others. The architecture is designed for permissioned, orchestrated machine-speed transactions, including microtransactions and stablecoin-supported payment types (Mastercard)
Adyen introduces Adyen Agentic for enterprise agentic checkout: Adyen launched a modular agentic-commerce suite built around Agentic Feed, Cart, and Payments APIs, with limited U.S. availability and compatibility with Meta AI checkout. The company framed the product as a way for enterprises to sell through conversational AI channels without rebuilding for each interface (Adyen)
Stripe previews Agentic Commerce Suite for UK businesses: Stripe said its Agentic Commerce Suite will begin rolling out later in 2026, enabling UK businesses to sell inside AI interfaces through a single integration. Stripe also noted that JD Sports and Wolf & Badger were already selling in U.S. AI surfaces through U.S. entities (Stripe)
Shopify opens UCP and Catalog API access to developers: Shopify made Universal Cart Protocol and Catalog API access self-serve, allowing developers to register an agent profile and call a public MCP endpoint (Shopify)
Arcade.dev raises $60 million for secure agent action infrastructure: Arcade.dev raised a $60 million Series A led by SYN Ventures, bringing total funding to $72 million. The company is building an authorization, reliability, governance, and audit layer for production AI agents that need to take real-world actions across enterprise systems (Business Wire)
Decagon introduces Duet Autopilot for self-improving customer experience agents: Decagon announced Duet Autopilot, a system designed to diagnose and improve AI customer-service agents against verified evaluation tasks. The company said its internal benchmark, DuetBench, showed Autopilot passing 93% of diagnostic tasks, above the average human score (Business Wire)
WILLY CHAVARRIA debuts World Cup collection through Swap’s AI storefront: Swap and WILLY CHAVARRIA launched an exclusive fashion collection through an AI-powered storefront with guided discovery, personalized recommendations, virtual try-on, and checkout inside the branded experience. The release is a useful example of agentic commerce moving from infrastructure demos into vertical retail moments (Business Wire)
HSBC pilots B2B agentic payments in Singapore with Mastercard: HSBC and Mastercard tested agent-led B2B payment flows in Singapore, extending agentic commerce into corporate procurement rather than consumer checkout alone. HSBC also paired the effort with mobile virtual cards that allow corporate clients to set spend limits, usage rules, and assignments by team, supplier, or project (FinTech News Singapore)
Ripple releases XRP Ledger AI Starter Kit for agentic payments: Ripple launched a developer toolkit for building agentic payment applications on the XRP Ledger, positioning XRPL’s settlement speed, predictable fees, and native payment functions as primitives for machine-to-machine commerce (Ripple)
Fastly and Skyfire enable trusted agentic commerce at the edge: Fastly detailed an integration with Skyfire that uses Know Your Agent credentials, JWT-based identity, and payment signals to let merchants verify agent traffic and enforce catalog, content, and checkout policies at the edge (Fastly)
Adobe introduces Adobe Brand Visibility for AI-discovery optimization: Adobe launched a product for measuring how brands appear across AI-generated answers, tracking share of voice across systems such as ChatGPT, Claude, Perplexity, Google AI Mode, and Copilot, and tying AI visibility to analytics and revenue attribution (Adobe)
ShopAgentic raises €1.9 million for agentic-commerce infrastructure: Hannover-based ShopAgentic closed an oversubscribed pre-seed round led by May Ventures and Greenfield Capital to build merchant-side infrastructure for AI-agent-driven product discovery, pricing, customer service, fulfillment, and checkout (ShopAgentic)
EasyCommerce 1.40 adds autonomous AI agents to WordPress ecommerce: EasyCommerce released two embedded agents for WordPress stores: a shopper-facing agent that can search catalogs, check availability, apply coupons, create orders, and generate payment links, and an admin copilot for store-owner workflows (EasyCommerce)
SAP and Google Cloud expand autonomous CX architecture for enterprise commerce: SAP said its CX organization and Google Cloud are building a real-time customer-experience foundation that connects trusted business context, AI agents, engagement channels, and commerce execution (SAP)
Sapiom acquires Fewsats to add machine-native payments to agent execution infrastructure: Sapiom acquired Fewsats, a payment-infrastructure company for AI agents, bringing founder Jordi Montes into the company to fuse payments into Sapiom’s agent runtime (Sapiom)
Pacvue launches Pacvue Prism, its “Agentic Commerce Grid” for commerce media orchestration: Pacvue introduced Prism at Cannes, positioning it as a commerce-native platform connecting retail media, search, programmatic, social, shoppable formats, ChatGPT integrations, The Trade Desk, Pinterest, tvScientific, Firework, Shopsense AI, and future Yahoo DSP integrations. The more interesting read is that agentic commerce is moving into media planning and measurement: brands are trying to connect AI discovery, retail media, and conversion signals into one operating layer (Pacvue)
DaVinci Commerce launches Agentic BrandStore Enterprise for AI-platform commerce: DaVinci Commerce announced general availability of Agentic BrandStore Enterprise, an AI-commerce experience layer designed to cover discovery, experience, and purchase across AI platforms. The platform is aimed at brands and retailers trying to make product data, creative, and purchase flows usable inside AI-mediated journeys rather than only on owned sites (PRNewswire)
MoonPay acquires Entendre to add AI finance agents to stablecoin payment infrastructure: MoonPay acquired Entendre, a company building AI accounting agents for high-volume, multi-entity, multi-currency businesses. This belongs in the agentic commerce stack because autonomous commerce does not end at payment authorization; reconciliation, treasury, close, audit, and ledger operations also need to become machine-operable if stablecoin and agent-led transaction volumes scale (PRNewswire)
Microsoft Advertising adds Product Explorer in Merchant Center: Microsoft introduced Product Explorer, giving retailers and advertisers a searchable view of catalog status, eligibility, metadata gaps, serving, and performance. This is directly relevant: machine-mediated discovery raises the value of clean, eligible, queryable product catalogs across ads, search, Copilot, and AI shopping surfaces (Microsoft Advertising)
L’Oréal and OpenAI bring Maybelline virtual try-on toward ChatGPT: L’Oréal said Maybelline New York will bring Makeup Virtual Try-On directly to ChatGPT using L’Oréal’s ModiFace technology, while L’Oréal works with OpenAI to strengthen product discovery for brands including Lancôme and Kérastase in the U.S. (L’Oréal)
Alchemy launches AgentCard with Visa Intelligent Commerce integration: Alchemy introduced AgentCard, a payments and identity platform for AI agents built on Visa Intelligent Commerce, provisioning agents with Visa payment tokens, dedicated email addresses, phone numbers, crypto wallets, spend controls, merchant-category restrictions, transaction limits, and budgets. This is one of the cleaner examples of “agent identity + payment credential + operating credentials” being packaged as developer infrastructure (PRNewswire)
🛡️ Security & Fraud
Kasada named a Leader in bot and agent trust management: Kasada was named a Leader in The Forrester Wave: Bot and Agent Trust Management Software, Q2 2026, with cited strengths across AI agent trust management, transaction assurance, protection, and bot/agent detection explainability (Business Wire)
Stripe extends Radar to AI-era abuse patterns: Stripe said Radar is expanding beyond card fraud to address multi-account abuse, free-trial fraud, pay-as-you-go abuse, and fraud across local payment methods such as Bacs (Stripe)
MRC frames agentic commerce as a digital-trust problem for merchants: The Merchant Risk Council’s June 10 agentic-commerce webinar focused on how merchants can prepare for AI agents that search, compare, and make purchase decisions on behalf of consumers. The emphasis on trust readiness reflects the operational gap between accepting agent-driven orders and reliably attributing intent, permission, and liability (Merchant Risk Council)
Affinidi and CardInfoLink deploy an AI-agent trust layer on Agenzo: CardInfoLink deployed Affinidi’s Agent Gateway on Agenzo, its agentic-commerce platform for travel and hospitality merchants, creating governed and auditable channels for AI agents connecting to merchants (PRNewswire)
DataDome named a Leader in bot and agent trust management: DataDome said it received the highest Current Offering score in The Forrester Wave: Bot and Agent Trust Management Software, Q2 2026, including top scores in AI agent trust management, intent visibility, web and LLM scraping management, and transaction assurance (DataDome)
HUMAN named a Leader in bot and agent trust management: HUMAN said its Human Defense Platform, including AgenticTrust, received top scores across criteria including AI Agent Trust Management, Threat Research, Intent Visibility, Attack and User Analytics, and Security Operations Integrations (HUMAN)
Fastly says AI traffic is growing 6.5 times faster than human traffic: Fastly reported that AI requests across its network grew roughly 30% from January through May 2026, with autonomous machine-to-machine traffic approaching half of internet requests (Fastly)
Wibmo unveils ARIA for financial-crime operations: Wibmo introduced Agentic Risk Intelligence Assistant for fraud, AML, KYC, and dispute operations, using specialized AI agents for signal aggregation, investigation support, recommendation drafting, and auditability while keeping critical decisions under human control (ANTARA News)
BCG warns agentic AI could industrialize financial scams: BCG argued that agentic systems could reduce the cost of running scams and fraud by 90% or more within two years, potentially doubling successful scam and fraud activity if banks and payment firms do not build adaptive AI-enabled defenses (BCG)
Fime frames intent verification as the missing control in agentic commerce: Fime’s FACT framework is positioned around checking whether an AI agent’s requested transaction still matches the user’s original mandate at the moment of purchase, rather than relying only on onboarding or payment authentication (FinTech News Hong Kong)
📈 Consumer & Market Insights
Checkout.com quantifies the trust gap in AI shopping: Checkout.com reported that one-third of surveyed consumers expect at least 10% of their purchases to be AI-driven within a year, while 27% said they trust no organization to operate an AI shopping agent. Merchants also appear ahead of actual transaction volume: 89% said they are preparing for agentic payments, while only 3% of transactions currently involve AI agents (Checkout.com)
AI-referred shoppers show stronger commercial intent: Adobe Analytics data reported by Reuters found that U.S. shoppers referred by large language models in May generated 53% more revenue per visit, converted 54% higher, and spent more time on retail sites than other shoppers. The data suggests AI discovery may be developing into a qualified-consideration channel rather than a purely upper-funnel behavior (Reuters)
Sopra Steria sizes European agentic commerce at more than €310 billion over ten years: Sopra Steria published a European study of 8,400 consumers across eight countries, estimating that AI agents could assist more than €310 billion of European ecommerce transactions within a decade. The trust gap remains central: 41% of Europeans do not yet trust any single actor to provide a shopping agent, while banks rank as the most trusted provider category (PRNewswire)
Rezolve Ai says 70% of shoppers now buy after a single search: Rezolve Ai reported that 70% of purchases in its analyzed data now happen after one search, up from 60% in 2025, while long-form queries rose from 25% of searches in 2024 to 40% by February 2026 (Rezolve Ai)
PYMNTS and Visa Acceptance Solutions report merchant readiness gaps for AI-agent shopping: The June merchant edition of the Global Digital Shopping Index found that merchants expect AI agents to influence shopping, payments, and loyalty, but few can clearly identify AI-driven traffic and purchases today. The report also says 87% of merchants still see checkout as falling short (PYMNTS)
Adobe reports AI travel traffic is scaling quickly but still trails on conversion: Adobe said AI-sourced traffic to U.S. travel sites grew 194% year over year in May 2026 and 2,215% since October 2024, with AI-referred travelers spending 70% longer on site and showing 21% higher engagement. Conversion still lagged non-AI traffic by 28%, but the gap has narrowed substantially since tracking began. (Adobe)
Shopify data points to stronger AI-referred ecommerce conversion: Marketing Tech News reported Shopify data showing AI-referred shoppers converting nearly 50% higher than organic search visitors, with average order values 14% higher. More than half of AI-referred sessions reportedly began on product detail pages, compared with roughly one-fifth for organic search. (Marketing Tech News)
🎯 Strategic Hiring Highlights
Product / platform / architecture
Gap Inc. — Principal – Architecture, AI, Unified Commerce — San Francisco, CA — $209,700–$272,600/yr — Gap Inc. Careers
Amazon — Principal PM, Agentic AI Shopping, RBS — Bengaluru, India — Salary not listed — Amazon Jobs
Accenture — Agentic Commerce Manager | Comms, Media, & Technology — Multiple locations — Salary not listed — Accenture Careers
Accenture — Agentic Commerce Senior Manager | Consumer Goods & Retail — Multiple locations — Salary not listed — Accenture Careers
Accenture — Commerce Consultant (Digital and Agentic) — London, UK — Salary not listed — Accenture
Novartis — Product Manager, Agentic AI Solutions — Basel, Switzerland (or other Novartis hubs, hybrid) — Salary not listed — Novartis Careers
Engineering / agentic platforms & infra
Nordstrom — Principal Engineer: AI Agentic Platform — Seattle, WA (hybrid) — $191,000–$297,000/yr — Nordstrom Careers
Amazon — Sr. Product Manager – Technical, Agentic AI, AWS Transform — Seattle, WA — $152,200–$205,900/yr — Amazon Jobs
Accenture — Agentic Data Engineer — Prague, Czech Republic — Salary not listed — Accenture Careers
Accenture — Technical Commerce & AI Manager | Agentic Commerce & AI — Multiple US locations — Salary not listed — Accenture Careers
Hexion — Lead Agentic AI Engineer — US Remote (Columbus, OH hub) — Salary not disclosed — Hexion Careers
The Trade Desk — Staff Software Engineer – Agentic AI — London, UK — Salary not disclosed — The Trade Desk Careers
Shape Security (F5) — Sr. Software Engineer – Agentic AI — Bengaluru, India — Salary not disclosed — Shape Security Careers
Duffel — AI Engineer, Backend — Remote (location not specified) — Salary not disclosed — Duffel (Ashby)
Trust, payouts & agentic‑commerce operations
Handshake — Senior Software Engineer, Pay & Safety (AI worker payouts, fraud, and compliance infra) — San Francisco, CA — $225,000–$250,000/yr — Handshake (Ashby)
Agentic‑adjacent platform & research roles
SanDisk (Western Digital) — Agentic AI Intern – Consumer, Summer 2026 (MBA or PhD) — Milpitas, CA (remote/hybrid) — Paid internship — SanDisk / Western Digital Careers
Novartis — Product Manager, Agentic AI Solutions — Basel (or Novartis research hubs) — Salary not disclosed — Novartis Careers
📖 Articles Worth Reading
AI agents won’t transform commerce until retailers redesign how decisions get made: A practical operating-model piece arguing that retailers need clearer decision ownership, audit trails, and data foundations before autonomous commerce can move beyond experiments (TechRadar)
Everyone is fighting over the wrong part of agentic commerce: This essay makes the case that B2B procurement, structured product data, and machine-readable buying workflows may produce earlier agentic-commerce value than consumer-facing autonomous checkout (TechRadar)
What Is Agentic Commerce? The 2026 Guide: Fin’s guide is a concise read on how agent-mediated shopping changes support, discovery, and conversion workflows for commerce and customer-experience teams (Fin)
NRF APAC 2026: Agentic Commerce Is Exciting. The Agentic Operating Model Is More Important: Bain’s useful contrarian point is that agentic commerce is not primarily a front-end shopping shift; it is an operating-model problem involving decision rights, talent, agent governance, KPIs, and merchant-LLM partnerships. This is a stronger board-level framing than most checkout-centric coverage (Bain & Company)
Permission, Not Payments, Will Shape the Agentic Commerce Revolution: PYMNTS and Paymentology argue that payment execution is not the scarce layer; delegated authority, identity, authorization, and fraud controls are. The piece is worth reading because it reframes agentic commerce around who owns the decision layer before money moves (PYMNTS)
J.P. Morgan Payments’ Michael Lozanoff on why agentic commerce can’t scale without governance: Tearsheet’s interview pushes past “can agents buy?” and focuses on whether merchants, banks, and PSPs can govern agent identity, permissions, limits, exceptions, and audit trails. The sharper implication is that intelligence becomes table stakes while governance becomes the durable moat (Tearsheet)
WarGames, Shopping Bots, and the Statute Trap: The CFAA and Amazon v Perplexity: Truth on the Market treats the Amazon–Perplexity dispute as more than a platform-access fight, arguing that decades-old anti-hacking law is a poor fit for user-authorized AI agents (Truth on the Market)
AI shopping agents are coming. No one is ready for them: Fortune’s Brainstorm Tech coverage is valuable because it captures the unresolved commercial plumbing: retailer blocking, missing standards, weak product recommendations, fraud, refunds, returns, and perceived liability (Fortune)
Agents, Agents Everywhere: Designing for AI Buyers vs AI Coworkers: Automattic Design separates two ideas that are often collapsed together: AI agents buying from stores and AI agents operating for merchants. The distinction is operationally useful because each requires different UX, controls, permissions, and merchant mental models (Automattic Design)
Consent Capital: Abhi Yadav’s essay argues that consent is moving from compliance burden to strategic asset as agents transact under standing permissions. The contrarian edge is that first-party data may be less defensible than durable, revocable, machine-readable customer consent (Abhi Yadav)
‘Companies that can serve both human and agent audiences will be the ones that survive’: TechRadar’s interview with WordPress VIP CTO Brian Alvey is a strong read on why websites now need to serve humans and software agents simultaneously. The underappreciated point is that agent-readable content may improve classic SEO while preserving direct customer relationships against AI and social gatekeepers (TechRadar)
The agentic advertising economy: From attention to action: McKinsey connects agentic commerce to advertising-market structure, arguing that discovery, recommendation, transaction, and measurement may consolidate inside AI-native environments and walled gardens. This is a useful counterweight to merchant-centric narratives because it shows where ad value could migrate when agents compress the funnel (McKinsey)
VISTA: A Versatile Interactive User Simulation Toolkit for Agent Evaluation: VISTA addresses a core blind spot in agent evaluation: static benchmarks miss multi-step user behavior and failure modes. Its ecommerce-shopping and customer-service settings make it relevant for teams trying to test agents before exposing them to live merchants, shoppers, or support queues (arXiv)
Agentic Commerce: How to Prepare Your Store When Customers Shop Through AI Agents: Origami’s guide is practical for merchants because it treats agentic commerce as operational readiness: structured product data, policy clarity, checkout reliability, and machine-readable store information (Origami)
Agentic commerce is scaling faster than its legal infrastructure: Proskauer’s June 22 analysis is worth reading alongside Amazon–Perplexity because it treats agentic commerce as a contract, liability, disclosure, agency, and regulatory problem rather than a UX upgrade. The key point is that delegation creates legal ambiguity before it creates operational scale (Proskauer)
McKinsey at Cannes Lions 2026: How AI is rewiring growth: McKinsey’s Cannes package is a useful strategy read because it links agentic commerce to advertising, commerce media, growth workflows, and full-funnel measurement. The unexpected angle is that the agentic-commerce battleground may show up first in media orchestration and budget allocation, not embedded checkout (McKinsey)
🧭 Looking Ahead
CommerceNext Growth Show 2026
Date: June 23-24, 2026
Location: New York, NY
Focus: Ecommerce growth, retail media, digital commerce operations, AI-enabled customer acquisition, and retention.
GITEX AI Europe 2026
Date: June 30-July 1, 2026
Location: Berlin, Germany
Focus: Enterprise AI, cybersecurity, cloud infrastructure, AI governance, and cross-sector technology deployment.
Stripe Tour Berlin
Date: June 30, 2026
Location: Estrel Berlin
Focus: Stripe product updates, global payments, AI-economy commerce infrastructure.
Berkeley Agentic AI Summit 2026
Date: August 1-2, 2026
Location: Berkeley, CA
Focus: Agentic AI research, infrastructure, interoperability, governance, and academic-to-industry transfer
Stripe Tour Sydney
Date: August 19, 2026
Location: ICC Sydney
Focus: Payments, software platforms, AI-commerce tooling, and internet-economy growth.
Stripe Tour Singapore
Date: August 25, 2026
Location: Sands Expo and Convention Centre
Focus: Regional payments, cross-border commerce, AI-enabled business infrastructure.
MRC San Diego 2026
Date: September 14–16, 2026
Location: Hyatt Regency Mission Bay Spa and Marina, San Diego
Focus: Payments, fraud prevention, chargebacks, merchant risk, and digital-trust operations.
Agentic Commerce & Payments Summit
Date: September 15, 2026
Location: Stockholm, Sweden
Focus:Agentic commerce, AI-driven checkout, identity, fraud, wallets, embedded finance, and payment orchestration.
General information only. Not legal, tax, investment, or professional advice. No warranty as to accuracy or completeness. Verify independently and consult your own advisers.
If you believe any information is inaccurate, please contact AgenticCommerce@proton.me and we will make a good-faith effort to review and correct it where appropriate.
Sponsored or affiliate content, if any, is disclosed.