The Agentic Commerce Frontier 📅 | March 10 – March 17 2026

Thanks for subscribing. This issue covers March 10–17, 2026, and the clearest signal from the week is that agentic commerce kept moving deeper into the stack: merchant feed infrastructure, payment delegation, enterprise orchestration, and trust controls all advanced faster than glossy consumer storefront headlines did.

This week my primer focuses on the subtle signals that may appear in Merchant logs and data that are tell tale signs that Agents haven’t waited for you to be Agent-Ready, to come visit, and what you should do about it.

---

🔥 TL;DR

• Mirakl and J.P. Morgan Payments moved agentic commerce closer to enterprise deployment by tying catalog orchestration to payment, tokenization, and fraud infrastructure

  • My takeaway: Wiring this stuff together is what will move the dial

• Santander and Visa completed controlled pilot agentic-commerce payments across Argentina, Brazil, Chile, Mexico, and Uruguay, showing AI-assisted purchasing is moving from concept to regional execution

  • My takeaway: A lot of the Agentic Commerce news is U.S.-centric, always good to see that things are progressing in other regions. I suspect Agentic Commerce will look quite different depending on regions, particularly on payment rails

• Amazon expanded Shop Direct through third-party product feeds, widening merchant access to Amazon discovery while preserving merchant-of-record flows and supporting Buy for Me purchases

  • My takeaway: Amazon wants agentic commerce on its own terms: block third-party agents from accessing Amazon, while extending Amazon-powered discovery and buying into other merchants’ ecosystems. The next fight in commerce is not just over transactions, but over who gets to be the shopper’s default agent

• Lemrock raised €6 million ($7 million) to build middleware that helps brands sell through AI agents like ChatGPT and Claude, positioning itself between merchant systems and conversational interfaces

  • My takeaway: Investors (rightly) believe there is a real opportunity in making existing commerce stacks legible and executable for AI agents

• Azoma launched the Agentic Merchant Protocol (AMP), a merchant-side standard designed to help brands control how their catalogs, brand rules, and product intelligence are represented across AI agents, with early adopters including Mars, L’Oréal, Unilever, Beiersdorf, and Reckitt

  • My takeaway: Yet another indicator that brands do not want to cede representation, ranking, and compliance logic to platform-owned protocols

---

🤖 Agentic Commerce Primer: The Real Subtle Signals Your Funnel Is Being Rewired by AI Agents

TL;DR: Inside your own systems, the real signal is when small, AI‑tainted cohorts outperform everything; Direct stops being truly direct; your catalog gets hammered by structured bursts; feed quality outguns UX; and payments data starts talking about delegation.

When those internal signals line up, your funnel is already agentic. The question is whether you treat that as luck, or as a design constraint for your next phase.

Who remembers Lost, the TV series where Oceanic Flight 815 crashes on a mysterious island and the survivors start noticing strange, unexplained phenomena?

In Lost, the survivors don’t figure out the island is strange because someone hands them a map. They figure it out because the details stop making sense. A polar bear on a tropical beach. A radio message looping for years. Numbers that keep showing up where they shouldn’t.

Agentic commerce is arriving the same way.

You will not wake up one morning with a new channel in your dashboard called “AI agents.” Instead, you’ll see tiny pockets of traffic that convert absurdly well, “Direct” visits that behave like pre‑qualified referrals, and machine‑like bursts against your product endpoints. Your own data starts behaving like that polar bear: it doesn’t fit the story you thought you were in.

---

1. Tiny AI cohorts, outsized performance

Adobe finds that AI‑driven visitors convert about 31% higher and generate roughly 254% more revenue per visit than other traffic, while viewing more pages and staying longer. Stripe and HUMAN see the same pattern: AI‑influenced traffic is a small slice of sessions but heavily over‑indexed on revenue and intent.

Inside your analytics, that looks like:

• Small, clearly AI‑origin cohorts (AI search, assistant links, chat plugins) doing 2–3× the revenue per session of paid search

• Very high conversion and AOV off relatively low volume

If a tiny segment is quietly your best “channel,” that’s your first sign an upstream agent is pre‑qualifying demand for you.

---

2. Journeys become product‑dense, not necessarily shorter

The data does not consistently show fewer pageviews. It shows more focused pageviews.

Adobe and others report that AI‑origin visitors view more pages per session and spend 30–45% more time on site, yet still convert better. HUMAN sees about 87% of agent traffic hitting product pages versus a tiny fraction on account or checkout flows.

What shifts internally:

• Your best‑performing cohort over‑indexes on:

  • Product detail pages

  • Pricing and promotion content

  • Availability, shipping, and policy pages

• The same cohort under‑indexes on:

  • Home and brand pages

  • Generic navigation

  • Editorial/blog content

That is what it looks like when exploration and comparison are handled off‑site, and your site is now where decisions get confirmed and executed, not where they’re formed.

---

3. “Direct” starts acting like a mystery super‑channel

Referrers often get stripped in AI and zero‑click flows, so a lot of AI‑mediated traffic lands in “Direct/None.” BCG’s warning: this erodes direct as a signal of true brand demand and hides how much of your funnel is being shaped off‑site.

Internal warning signs:

• Direct sessions increasingly land deep on PDPs or carts, not your homepage

• Direct conversion and revenue per visit begin to resemble your AI cohort more than your organic baseline

• Direct revenue spikes that don’t line up with any obvious brand, email, or offline activity

When “Direct” behaves like a highly optimized referral channel with no corresponding campaign story, assume a chunk of it is AI mediation until you’ve proven otherwise.

---

4. Agentic “bursts” against product, price, and inventory

At the infrastructure level, agents leave clear fingerprints.

Stripe calls out “agentic bursts”: short windows where an agent issues dozens of product or stock checks to your endpoints in seconds. HUMAN’s telemetry shows most agent traffic is concentrated on product surfaces.

In your logs and metrics, this appears as:

• Tight spikes of GETs to product details, pricing/promos, inventory, and shipping APIs.

• Highly regular patterns (walking category trees, hitting sequential SKUs, pulling slices of feeds).

• Activity at machine speed from a small cluster of IPs or user agents, not broad human‑like dispersion.

If those bursts correlate over time with downstream high‑intent sessions or orders, your catalog is already being used as structured input to someone else’s ranking and recommendation logic.

---

5. Structured data begins to outperform UX as a growth lever

Adobe, Stripe, and others are converging on the same point: for a growing slice of traffic, agents are the real consumers of your catalog, and they care far more about feeds and schema than front‑end design.

You’ll feel this internally when:

• Fixing product feeds, schema markup, prices, and inventory accuracy moves AI‑influenced performance more than redesigning templates or rewriting copy

• Cleaning structured data (including agent‑targeted catalog feeds) coincides with:

  • Better behavior from AI‑tagged cohorts

  • Fewer “wrong product / wrong price” tickets from customers arriving via assistants

At that point, your main controllable surface for agentic demand is not the “experience layer.” It’s the quality and completeness of your machine‑readable catalog.

---

6. Payments and risk telemetry pick up “delegated intent”

You don’t need external announcements to see this, your own payments and fraud data will change.

Mastercard’s Verifiable Intent and Stripe’s Shared Payment Tokens exist specifically to support agent‑initiated transactions with proof of user delegation. As they roll into your stack, signals inside your systems include:

• New token types or flags showing up in payment logs that denote agent‑initiated or delegated transactions

• Fraud and risk reviews that hinge on “what did the user authorize the assistant to do?” rather than just “did this card authenticate?”

• Emerging dispute patterns where customers accept that something was authorized but contest the scope or timing (“I approved research, not purchase”)

Those are internal signs that agents are no longer just browsing, they’re starting to execute transactions, and your trust model is adapting around that.

---

What should you do about these signals?

There are a few things you can do in the next 90 days:

• Tag and track AI‑influenced and AI‑suspect cohorts separately

• Instrument your logs for agent‑like bursts on product, price, and inventory endpoints

• Treat catalog data quality (feeds, schema, APIs) as a first‑class KPI, not back‑office hygiene

• Review payments and disputes for early signs of delegated intent and new token types

There are tactical upgrades to your instruments and plumbing. They do not replace the harder work: deciding what your Agentic Commerce strategy is, whether you want to be just a reliable endpoint in other people’s agents, or build your own agent, your own rails, or your own leverage in the ecosystem.

But until you make those calls, you at least want to know when there’s a polar bear on your beach.

---

🚀 Major Announcements & Funding News

• Mirakl & J.P. Morgan Payments enable agentic commerce at enterprise scale: On March 10, Mirakl and J.P. Morgan Payments announced a global agreement that combines Mirakl Nexus with J.P. Morgan’s payment infrastructure so merchants can support secure, large-scale AI-agent transactions without stitching together their own commerce and payments stack (Mirakl)

• Santander and Visa complete Latin America’s first end-to-end AI-agent payments pilot: Santander says the controlled pilot ran across five Latin American markets and used Visa Intelligent Commerce to let consumers securely delegate shopping tasks to trusted AI agents (Santander)

• Amazon broadens Shop Direct through feed-based merchant onboarding: Amazon said merchants can now connect catalogs through Feedonomics, Salsify, and CEDCommerce, bringing more sellers into Shop Direct and extending discovery across Amazon search and Rufus, while Buy for Me lets Amazon complete eligible purchases on the shopper’s behalf (Amazon)

• Azoma launches AMP to give brands more control in agentic commerce: Azoma introduced Agentic Merchant Protocol (AMP) as a merchant-side layer for catalog control, multi-agent syndication, and brand-safe representation, with early adopters including Mars, L’Oréal, Unilever, Beiersdorf, and Reckitt (Business Wire)

• Mastercard pushes agentic AI into SMB decisioning with Virtual C-Suite: Mastercard unveiled Virtual C-Suite, with Virtual CFO as the first planned module, positioning AI as an operating layer for small-business finance decisions rather than just a reporting assistant (Mastercard)

• Fareway selects Instacart Storefront Pro for its upgraded digital grocery experience: Fareway’s launch adds pickup on the Instacart Marketplace, no-markup ordering, and an updated digital storefront built on Instacart’s enterprise stack, which includes AI-powered search and merchandising (PR Newswire)

• Constructor adds more scale and agentic tooling to product discovery: Constructor reported 82% customer growth, 322 billion product-discovery interactions, and rollout momentum for AI Product Insights Agent and AI Shopping Agent, reinforcing how search/discovery vendors are repositioning around agent-led journeys (PR Newswire)

• Wizard partners with Stripe on ACP for retailer purchases: Wizard said that it is joining Stripe’s early ACP partner group and will use Stripe’s infrastructure to help retailers turn AI-driven discovery into completed purchases while keeping control of brand, catalog, fulfillment, and customer relationships (Business Wire)

• Lemrock raises €6 million ($7 million) to build agentic commerce infrastructure: Paris-based startup Lemrock announced on March 11 that it raised funding to help brands sell inside AI agents such as ChatGPT and Claude, positioning itself as a middleware layer between merchant systems and conversational interfaces (EU-Startups)

• Zendesk announces proposed acquisition of Forethought: Zendesk said that it entered into a definitive agreement to acquire Forethought, aiming to expand self-improving AI agents across customer-service workflows and accelerate its roadmap for agentic support across channels (PR Newswire)

• Shoplazza adopts agentic commerce architecture to power AI-driven ecommerce operations: Shoplazza announced that it is rolling out an agentic commerce architecture across its platform, introducing AI agents that can directly execute operational ecommerce tasks for DTC brands (PR Newswire)

---

🛡️ Security & Fraud

• Nasdaq Verafin says financial crime hit new highs in 2025: Verafin’s report puts illicit financial activity at $4.4 trillion and says fraud losses exceeded half a trillion dollars worldwide, underscoring how the trust layer is becoming central to agent-enabled payments (Verafin)

• Payments Dive: AI is materially worsening the fraud environment: Reporting on the Verafin findings, Payments Dive said fraud and scam losses rose 9.2% in 2025, with much of the increase attributed to criminals using AI (Payments Dive)

• UK fraud data shows AI-enabled scams are spilling into shopping and account abuse: The Guardian reported that the UK saw a record 444,000 fraud cases in 2025, with Cifas pointing to AI-enabled impersonation, account takeover, and scams spanning mobile, online shopping, and credit accounts (The Guardian)

• The UK launched a new 2026–2029 fraud strategy: The Home Office said the strategy is backed by more than £250 million of investment and is meant to create a more system-wide response to fraud affecting people and businesses (GOV.UK)

• Amazon wins court order blocking Perplexity’s AI shopping agent: Reuters reported on March 10 that a federal judge temporarily barred Perplexity from using its shopping agent on Amazon after Amazon argued the tool unlawfully accessed customer accounts, making this a consequential trust and authorization case for agent-led commerce (Reuters)

• Agentic AI and consumer law: the CMA’s guidance for businesses: Lewis Silkin’s note is worth reading because it translates the UK CMA’s position into practical implications for businesses using AI agents in marketing, refunds, and customer interactions, stressing that firms remain responsible for what their agents do (Lewis Silkin)

• Commerce agentique : qu’est-ce que ça change ?: Stratégies’ feature is a useful French-language overview of how LLM-based discovery, visibility battles, product-data quality, and middleware providers are reshaping how brands compete in agentic commerce (Stratégies)

• Enterprises race to build “Know Your Agent” defenses against AI bots: A new PYMNTS Intelligence and Trulioo framework says nearly 90% of organizations now view bot activity management as a challenge, and argues that agentic commerce will require a five-layer verification model that extends KYC and KYB to autonomous agents. (PYMNTS)

  ---

  📈 Consumer & Market Insights

• Enterprise commerce has moved from AI pilots to budgeting and integration work: Across the latest Logicbroker and Digital Commerce 360 reporting, surveyed enterprises show high deployment and spend intent: 95.5% report at least one AI capability already live, 47.3% plan to invest at least $1 million over the next year, 90.7% think AI will influence at least 20% of ecommerce orders by 2027, and many now see systems integration as the real blocker (Digital Commerce 360)

• EU online shopping still has a friction problem: Eurostat said 35.4% of EU online shoppers encountered problems in 2025, with slow delivery (19.9%), poor site usability (11.5%), and incorrect or damaged goods (10.4%) as the top complaints. For agentic commerce, that is a reminder that better discovery will not matter if execution stays messy (Eurostat)

• Constructor’s scale numbers show where commerce AI is already monetizing: Beyond the headline growth, Constructor said its AI Product Insights Agent has produced retailer gains including 52% higher add-to-cart rates and 56% higher purchase rates, suggesting that merchant-controlled AI guidance is already paying off at the PDP layer (PR Newswire)

• One to One Monaco shows agentic AI becoming a mainstream European retail priority: EcommerceMag France’s coverage of One to One Monaco says agentic AI was central to the event’s framing of commerce’s “new codes,” with Fevad data showing improving retailer sentiment, rising concern over Chinese competition, and broad recognition that AI is reshaping the sector’s operating model (EcommerceMag)

• Acquirers say payments rails are ready for AI shopping agents, but the market is not: A new PYMNTS Intelligence report commissioned by Visa Acceptance Solutions finds that many acquirers believe existing payments infrastructure can support agent-led commerce, but say adoption still depends on solving merchant integration costs, legacy-system friction, and trust issues around fraud, identity, regulation, and liability (PYMNTS)

• Target is investing in agentic commerce as part of its turnaround plan: The Wall Street Journal reported on March 16 that Target’s tech leadership is investing in agentic commerce partnerships with companies like OpenAI and Google so shoppers can buy directly through AI platforms, signaling that large retailers now see agent-led purchasing as part of mainstream digital strategy rather than an experimental edge case (The Wall Street Journal)

  ---

  🎯 Strategic Hiring Highlights

• U.S. Bank — Senior Product Manager, Developer Experience & Agentic Commerce Enablement — Available in 7 U.S. locations / Hybrid — Salary not listed — U.S. Bank Careers

• Worldpay — Product Manager, Agentic Commerce — San Francisco, CA; Denver, CO; Atlanta, GA; New York, NY; Boston, MA — $130,000–$150,000 — Worldpay Careers

• Borderless — GenAI & Agentic Commerce Product Manager — London, UK (On-site, Visa Sponsorship) — Salary not listed — Borderless Jobs

• Trove — Product Manager, Sales Channel Platform (including agentic commerce channels) — Remote (US-based) — $130,000–$160,000 + equity — Welcome to the Jungle

• Stripe — Senior Staff Product Designer, Agentic Commerce — San Francisco, CA — Salary not listed — Stripe Jobs

• Stripe — Business Development Manager, Agentic Commerce — South San Francisco HQ or Seattle — $310,100–$465,100 OTE — Stripe Jobs

• Swap — Head of Product, Agentic Commerce — London or Amsterdam or Remote EU — Salary not listed — We Work Remotely

  ---

📖 Articles Worth Reading

• 10 things we learned building for the first generation of agentic commerce: Stripe’s latest field notes are useful because they focus less on hype and more on practical bottlenecks sellers are already hitting as agentic checkout moves into production environments (Stripe)

• AI Wants the Checkout but Spreedly Says Merchants Keep the Keys: PYMNTS captures a key strategic tension in agentic commerce: whether AI intermediaries control checkout, or whether merchants remain the merchant of record while orchestration and credential control sit underneath (PYMNTS)

• Why OpenAI’s Checkout Retreat Spells Trouble for Its Commerce Strategy: Jason Goldberg’s Forbes piece is worth reading as a framing argument about platform control, distribution power, and why commerce infrastructure may matter more than a flashy embedded checkout button (Forbes)

• Amazon expands a program that lets customers shop from other retailers’ sites: TechCrunch’s coverage is a concise read on why Amazon’s Shop Direct expansion matters beyond Amazon itself: it normalizes marketplace-adjacent discovery with merchant-site completion (TechCrunch)

• Shopify says purchases are coming ‘inside ChatGPT’ through agentic storefronts as OpenAI retreats on Instant Checkout: Even though it is member-only, the indexed summary is important because it points to merchant-native storefront models becoming more central inside chat interfaces (Modern Retail)

• Analyzing first-party fraud trends: account, free trial, and refund abuse: Stripe’s fraud post is a useful companion read for anyone thinking about autonomous shopping, because it shows how trust and abuse patterns are already shifting before agentic commerce hits mass scale (Stripe)

• Agentic commerce faces a reality check in B2B ecommerce: Digital Commerce 360 argues that enterprise agentic commerce is likely to sit on top of existing procurement, ERP, and ecommerce systems rather than replace them, because negotiated pricing, governance, and workflow complexity make full autonomy hard to deploy at scale (Digital Commerce 360)

• Which retailers are embracing agentic commerce?: Useful scan of early retailer adoption and experimentation, making it a good read for tracking who is moving first and how agentic shopping is being operationalized in practice (Retail Week)

• Agentic commerce 2026: AI agents are transforming shopping: Worth reading for its infrastructure framing, especially the argument that most retailers do not need a full replatform but do need a translation layer that makes product data, APIs, and workflows legible to AI agents (Invisible Technologies)

• Fashion executives are preparing for a shift whose scale is still uncertain: FashionNetwork’s article captures how fashion and luxury leaders are thinking about agentic commerce as a structural change in shopping behavior rather than just another interface experiment (FashionNetwork)

• Why agentic commerce is the new shopping channel: In a techUK guest essay, Shopify’s EMEA lead argues that merchants should treat agentic commerce as a new channel built on machine-readable catalogs, portable checkout, and open standards such as UCP, with merchant-of-record control preserved underneath (techUK)

• AI agents are learning to handle real-world products: Mastercard’s feature on Pentatonic is worth reading because it shows agentic commerce moving beyond digital catalogs into physical retail and circular commerce, with infrastructure for AI to identify, value, and transact on real-world goods in stores (Mastercard)

---

🧭 Looking Ahead

Shoptalk Spring 2026

• Date: March 24–26, 2026

• Location: Mandalay Bay, Las Vegas

• Focus: Shoptalk’s 2026 spring event is explicitly centered on retail in the age of AI. For agentic commerce operators, it is one of the clearest near-term venues to watch how retailers are reframing discovery, loyalty, merchandising, and fulfillment for AI-mediated shopping journeys.

eTail Asia 2026

• Date: March 24–25, 2026

• Location: Equarius Hotel, Sentosa, Singapore

• Focus: eTail Asia is positioned around ecommerce and omnichannel leadership in APAC. It should be especially relevant for teams tracking how agentic journeys intersect with marketplace-heavy commerce, mobile-first shopper behavior, and cross-border retail execution in Asia.

Chatbot Summit Berlin 2026

• Date: March 25, 2026

• Location: Berlin, Germany

• Focus: “Mastering Agentic AI,” with explicit “agentic commerce in action” themes and implementation learnings

Agentic Commerce Summit 2026

• Date: April 13–15, 2026

• Location: London, Europe

• Focus: A closed-door summit centered on autonomous AI agents in retail, agent payments, commerce protocols, multi-agent orchestration, and governance. The published agenda positioning makes it one of the most directly relevant near-term events for agentic commerce operators across retail, payments, and platforms.

Syndigo Connect 2026 — AI Commerce & Agentic Product Experiences

• Dates: April 14–16, 2026

• Location: Bellagio, Las Vegas, USA

• Focus: Billed as “the premier conference for AI-driven product experiences and agentic commerce,” centered on AI-ready product data, retail media, and connected experiences that determine how agents discover, compare, and buy across channels

NexGen Retail Summit, London

• Dates: April 15–16, 2026

• Location: Hilton Hotel, London, UK

• Focus: Retail tech summit on AI’s transformation of commerce—sessions span “The $1 Trillion AI Retail Revolution,” predictive commerce, omnichannel, frictionless payments, and Responsible AI in Retail

Adobe Summit 2026 — Commerce programming (“Adobe Commerce Summit” track coverage)

• Date: April 19–22, 2026

• Location: Las Vegas + Online

• Focus: Adobe is explicitly framing sessions around agentic/generative AI; Commerce sessions and labs are a strong proxy for enterprise merchant readiness

Money20/20 Asia

• Date: Apr 21–23, 2026

• Location: Bangkok, Thailand

• Focus: Cross-border rails (licenses, FX, A2A), wallets, and agentic acceptance

Money20/20 Europe

• Date: June 2–4, 2026

• Location: RAI Amsterdam, Netherlands

• Focus: Money20/20 Europe’s 2026 themes include AI operating inside rewired financial ecosystems. This is the payments-side event to watch for tokenization, delegated payments, identity, compliance, and infrastructure that make agentic checkout viable at scale.

NRF 2026: Retail’s Big Show Asia Pacific

• Date: June 2–4, 2026

• Location: Marina Bay Sands, Singapore

• Focus: NRF APAC will bring regional retailers and tech providers together around the “connected, AI-native store” and cross-border commerce. It is a strong signal venue for how APAC retail is adapting merchandising, store tech, and digital journeys around AI.

Shoptalk Europe 2026

• Date: June 9–11, 2026

• Location: Fira Gran Via, Barcelona

• Focus: Shoptalk Europe’s 2026 agenda already includes sessions on AI readiness, shifting consumer priorities, and retail transformation. For agentic commerce watchers, it should be one of the best places to hear how European retailers are thinking about data readiness, orchestration, and AI-enabled growth.