Microsoft Copilot Adds ESW Agentic Commerce Checkout in US

ESW is moving to turn Microsoft Copilot from a product-recommendation surface into a transactional storefront, launching an agentic commerce service that connects AI-driven discovery with embedded checkout. Available initially in the United States, ESW Agentic Commerce is designed to expose participating brands’ catalogs to Copilot, preserve links to their existing ecommerce operations, and let shoppers complete purchases without abandoning the AI conversation. The announcement matters because it tackles the difficult part of AI shopping: not generating recommendations, but converting a conversational request into an accurate, secure, fulfillable order.

---

Background​

Online retail has repeatedly reorganized itself around new points of consumer attention. Brands first optimized desktop websites for search engines, then built mobile storefronts and apps, and later adapted their catalogs for marketplaces, social platforms, retail media networks, and voice assistants.
Agentic commerce represents the next stage of that progression. Instead of requiring a shopper to visit several websites, compare product pages, inspect policies, and manually assemble a basket, an AI assistant can interpret the shopper’s intent and coordinate multiple parts of the buying process.

From search keywords to conversational intent​

Traditional ecommerce discovery begins with keywords. A shopper searches for “waterproof hiking jacket,” receives links or product cards, and performs most of the evaluation independently.
An AI assistant can process a much richer request: “Find a lightweight waterproof hiking jacket under $200 that will pack into a carry-on and arrive before Friday.” The agent must translate that sentence into structured criteria involving price, size, technical properties, availability, delivery location, and time.
That changes which product data matters. A conventional search feed may identify a jacket’s title, image, price, and category, while an AI agent needs reliable information about fit, materials, shipping eligibility, return rules, warnings, variants, and current inventory.

Microsoft’s expanding commerce ambitions​

Microsoft has been building commerce capabilities around Copilot that combine conversational discovery with product cards and in-conversation purchasing. Copilot Checkout is intended to reduce the handoff between an AI recommendation and the merchant’s transaction systems, allowing eligible shoppers to move from intent to payment without opening a separate storefront.
The company has also been developing a broader retail stack around Microsoft Merchant Center, merchant feeds, Copilot Studio, Dynamics 365 Commerce, and agent-oriented protocols. ESW’s arrival adds an international ecommerce operator that already handles operational functions such as payments, duties, localization, compliance, delivery, and returns for brands selling across borders.

Why ESW’s role is significant​

ESW is not positioning its new product as another consumer chatbot. It is presenting the Agentic Hub as an infrastructure layer that makes brand commerce capabilities usable by external AI agents.
That distinction is important. The market already has many systems capable of discussing products, but fewer can reliably confirm a variant, calculate the final price, process a payment, create an order, and coordinate fulfillment across jurisdictions.

---

What ESW Agentic Commerce Actually Does​

At a high level, ESW Agentic Commerce acts as a bridge among brand catalogs, AI shopping interfaces, and the operational systems required to complete an order. Microsoft Copilot is planned as the first integration, while ESW says additional agents and touchpoints will follow.
The service is meant to work alongside a retailer’s existing ecommerce infrastructure rather than force a costly replatforming project. That makes the proposition more practical for enterprise brands that have spent years integrating storefronts with product information, inventory, order management, customer service, logistics, tax, and financial systems.

Discovery is only the first layer​

Making a catalog available to an AI platform does not guarantee that the right product will appear. Product records must be structured, current, sufficiently detailed, and understandable to a system that may compare them against items from many competing merchants.
ESW says it will help brands integrate and optimize their catalogs for AI platforms. In practice, this category of work can include normalizing product attributes, exposing variant relationships, mapping categories, improving descriptions, supplying policy information, and maintaining consistency between the AI-facing feed and the source ecommerce system.
An agent also needs answers to questions that static marketing copy frequently ignores. It may need to know whether a charger supports a particular voltage, whether a garment contains wool, whether a cosmetic can be shipped to a location, or whether a product can arrive by a specified date.

Transaction execution is the harder problem​

Once a user accepts a recommendation, the system must turn a conversational selection into a valid order. That can require real-time checks for inventory, current pricing, promotional eligibility, shipping options, taxes, address restrictions, payment authorization, fraud signals, and order confirmation.
A recommendation based on stale data is inconvenient. A completed transaction based on stale data creates customer-service costs, refund requests, payment disputes, and reputational damage.
The defining feature of ESW’s announcement is therefore not that Copilot can describe products, but that ESW intends to connect the conversation to executable commerce. Embedded checkout shortens the distance between interest and purchase while shifting the burden onto the underlying infrastructure to remain accurate.

---

The Agentic Hub as an Infrastructure Layer​

ESW describes the Agentic Hub as the foundation of its agentic commerce offering. Its purpose is to expose global commerce capabilities to AI agents in a reusable form, allowing integrations to expand beyond Microsoft Copilot without rebuilding the entire connection for every platform.
This resembles the evolution of headless commerce, in which the storefront was separated from the commerce engine. Agentic commerce takes that separation further: the customer-facing interface may belong to an AI platform, while catalog, checkout, payment, compliance, and fulfillment functions remain distributed across the merchant and its service providers.

A translation layer for agents and retailers​

Most enterprise commerce environments were designed around websites, apps, call centers, and marketplace feeds. Their APIs may be powerful, but they were not necessarily built for an autonomous agent that requests multiple pieces of information and expects structured, machine-readable responses.
The Agentic Hub can be understood as a translation and orchestration layer. It must make commerce functions accessible in a form that an agent can call while respecting the merchant’s business rules.
A successful implementation would need to distinguish among actions with very different consequences. Reading a product description is low risk, reserving inventory has operational implications, and charging a payment method requires strong authorization and auditability.

Why reusable connections matter​

The agent ecosystem is unlikely to consolidate around a single interface immediately. Microsoft Copilot, ChatGPT, Google’s Gemini experiences, Perplexity, marketplace assistants, payment-provider agents, retailer-owned agents, and other systems are competing to mediate purchasing intent.
Building separate bespoke integrations for every agent would recreate the fragmentation that merchants already face across marketplaces and advertising platforms. A reusable hub could reduce that burden, although its value will depend on how well it supports emerging standards and how much platform-specific customization remains necessary.

Protocols will shape the market​

Agentic commerce is developing alongside several proposed technical protocols intended to standardize how agents discover tools, communicate with other agents, retrieve product information, and initiate payments. The industry has not yet settled on a universally dominant architecture.
For retailers, protocol support is more than an engineering detail. It can determine whether a brand remains portable across AI platforms or becomes dependent on a proprietary connection controlled by one large technology provider.
ESW’s multi-agent plan is strategically sensible because it acknowledges that uncertainty. The real test will be whether the Agentic Hub provides genuine portability or simply becomes another intermediary through which retailers must route their data and transactions.

---

How a Copilot Purchase Could Work​

The precise shopper experience will vary by product, merchant, payment method, and Copilot surface. Even so, the transaction can be understood as a sequence of coordinated steps.

The likely journey from request to order​

• The shopper expresses an objective in Copilot. The request may include requirements involving price, brand, compatibility, delivery timing, materials, color, or intended use.
• Copilot interprets and structures the request. The assistant identifies product criteria, resolves ambiguities, and may ask follow-up questions before searching eligible catalog data.
• Participating products are evaluated. The system compares structured merchant information, product attributes, commercial signals, availability, and other relevant factors.
• Copilot presents a shortlist. The shopper can inspect product cards, request comparisons, modify constraints, or ask why a product was recommended.
• The system validates the selected item. Before checkout, the commerce layer should confirm the exact variant, price, stock status, shipping eligibility, and other transaction-critical details.
• Checkout occurs inside the AI experience. The shopper supplies or confirms the information required for payment, delivery, taxes, and applicable policies.
• The merchant’s operational systems receive the order. Inventory, fulfillment, confirmation, customer service, and returns then proceed through the connected commerce stack.

Every transition presents a possible failure point. The product may be described incorrectly, a size may be unavailable, a delivery estimate may change, a discount may not apply, or the shopper may misunderstand what the agent is authorized to do.

Fewer clicks, but more hidden complexity​

Embedded checkout can eliminate page loads, repeated searches, and abrupt context changes. That reduction in visible friction may increase conversion, particularly when the shopper has already described a specific need and received a plausible recommendation.
The complexity does not disappear, however. It moves behind the conversation, where merchants, ESW, Microsoft, payment partners, and logistics systems must coordinate rapidly enough to preserve the impression of a simple interaction.
This is a recurring pattern in computing. The easiest interfaces often depend on the most elaborate infrastructure.

---

Why Microsoft Copilot Is a Logical First Integration​

Microsoft gives ESW access to a broad consumer and enterprise ecosystem while offering a commerce program already oriented toward merchant participation. Copilot can appear across web and mobile experiences, and Microsoft has strong incentives to turn conversational engagement into measurable commercial activity.
For Microsoft, ESW broadens the range of retailers that can connect to Copilot Checkout, particularly brands with complicated international requirements. For ESW, Copilot provides a prominent launch partner and a live environment in which to prove that its hub can support transactions rather than demonstrations.

Microsoft’s advantage in merchant relationships​

Microsoft has decades of experience connecting advertisers and merchants to digital demand through Bing, Microsoft Advertising, and merchant feeds. Those relationships give it a route for onboarding catalogs and explaining how visibility inside Copilot relates to existing product-data practices.
The company also operates enterprise platforms used by retailers, including Azure, Dynamics 365, Power Platform, and Copilot Studio. That does not guarantee that merchants will adopt Copilot Checkout, but it gives Microsoft many points of entry into retail organizations.

Copilot becomes a commercial destination​

The strategic shift is subtle but substantial. A general-purpose AI assistant begins as a place to ask questions, but embedded checkout turns it into a transaction environment.
That creates new incentives around recommendation placement, merchant eligibility, sponsored visibility, product ranking, and user retention. If consumers can research and buy without leaving Copilot, Microsoft gains a stronger position between the shopper’s intent and the merchant’s storefront.

Windows users are part of the distribution story​

For WindowsForum readers, the relevance extends beyond ecommerce infrastructure. Copilot is woven into Microsoft’s broader consumer software strategy, and shopping is one more category through which the company can make the assistant a recurring part of everyday computing.
Microsoft must still earn that behavior. Users will not delegate purchasing simply because an assistant is available in the operating-system ecosystem, especially if recommendations feel promotional, inaccurate, or intrusive.

---

The Cross-Border Commerce Challenge​

ESW’s strongest differentiation may emerge when the shopper, brand, inventory, and delivery destination span different countries. Domestic checkout is already complex, but international commerce adds currencies, duties, product restrictions, localized payment preferences, consumer-protection rules, and reverse logistics.
ESW says its platform supports end-to-end international operations across more than 200 markets. That existing footprint gives the company a foundation for answering questions that a generic AI interface cannot safely improvise.

The price must mean what the shopper thinks it means​

A shopper presented with a product price may assume it reflects the amount that will ultimately be paid. Cross-border orders can challenge that expectation when duties, taxes, shipping fees, foreign-exchange costs, or carrier charges appear later.
An agentic checkout must clearly communicate the total cost and the assumptions behind it. A smooth conversational interface will not compensate for a surprise customs bill after delivery.
This requires more than currency conversion. The commerce layer may need to determine the destination, product classification, local thresholds, applicable taxes, and whether duties are collected at checkout.

Availability is jurisdiction-specific​

A product available on a retailer’s primary website may not be eligible for shipment to every country. Batteries, cosmetics, food, medical products, encryption technology, luxury goods, and other categories can face carrier, customs, licensing, or regulatory restrictions.
AI agents must not treat global inventory as universally purchasable inventory. The answer to “Can I buy this?” depends on where the shopper lives, which warehouse will fulfill the order, and what rules apply along the route.

Returns can expose the weakest link​

A fast AI-assisted purchase can create a slow and expensive return. International returns involve labels, customs documentation, local collection options, refund timing, duty recovery, product inspection, and decisions about whether the item should travel back across a border.
Brands must ensure that an agent presents return policies accurately before purchase. If the conversational experience emphasizes convenience while obscuring costly or restrictive return conditions, customer trust will erode quickly.

---

The Retailer’s Technical Workload​

The promise of “working alongside” existing ecommerce platforms will appeal to retailers, but it should not be mistaken for zero integration work. Enterprise systems contain inconsistent identifiers, duplicated product records, regional catalogs, custom promotions, and years of business-specific logic.
Agentic commerce exposes those weaknesses because an AI interface needs clean answers in real time. A website can sometimes conceal data problems through manual merchandising, while an autonomous agent will propagate whatever information its connected systems provide.

Product information becomes operational data​

Retailers often treat product descriptions as marketing content. In agentic commerce, descriptions and attributes influence automated reasoning and can affect whether the agent considers a product suitable.
Important data may include:

• Variant relationships must be explicit, so the agent does not confuse a product family with a purchasable size, color, storage capacity, or configuration.
• Compatibility claims must be structured and verified, particularly for electronics, accessories, automotive parts, and software.
• Inventory signals must be timely, because an agent should not recommend an item that cannot be purchased.
• Policy information must be machine-readable, including returns, warranties, delivery limitations, and customer-support channels.
• Hazards and restrictions must travel with the item, rather than remain buried in a webpage or PDF.
• Regional differences must be preserved, including localized assortments, pricing, promotions, and legal notices.

This is effectively a data-governance project. Brands that begin with incomplete or conflicting records may need substantial remediation before agentic sales perform reliably.

Real-time APIs become essential​

Catalog feeds are useful for discovery, but checkout requires current information. The system may need live calls to validate price, stock, shipping, and promotional conditions immediately before the shopper confirms an order.
Latency matters because a conversational experience feels broken if every validation takes several seconds. Reliability matters even more because a quick incorrect answer can be more damaging than a slower accurate one.
Retailers will need monitoring that distinguishes among failures originating in Copilot, ESW, the brand platform, a payment processor, a fraud service, an inventory system, or a carrier integration. Without end-to-end observability, support teams may struggle to explain why an apparently simple conversation did not produce an order.

---

Consumer Experience and Trust​

The strongest consumer argument for agentic commerce is convenience. Shoppers can describe what they need in ordinary language, refine the results conversationally, and complete a purchase without navigating a maze of storefront menus.
Convenience alone will not secure adoption. Users must understand why products appear, which company is selling them, what data is being shared, and when an AI recommendation crosses into a commercial placement.

Recommendation quality will define the channel​

A recommendation engine can optimize for many outcomes: relevance, predicted engagement, conversion probability, merchant quality, price, delivery speed, advertising revenue, or platform relationships. Those priorities do not always align.
If a shopper requests the “best” laptop, the assistant must decide what best means. The most suitable product for the stated requirements may not be the product with the strongest historical conversion rate or the highest commercial value to the platform.
Microsoft and its partners will need to make useful distinctions among organic recommendations, sponsored placements, and merchant-funded incentives. An AI answer can feel more authoritative than a conventional advertisement, increasing the responsibility to label commercial influence clearly.

Consent must remain visible​

An agent that can purchase products needs explicit boundaries. The user should know whether it is only researching, building a cart, reserving inventory, applying a stored payment method, or placing an order.
High-quality systems should make consequential steps unmistakable and reversible where possible. They should also produce durable confirmations that state the merchant, item, variant, final price, delivery destination, and return terms.

Accessibility could improve​

Conversational shopping may help users who find complex ecommerce navigation difficult, including people with some visual, motor, cognitive, or literacy-related barriers. A shopper could ask the assistant to explain technical differences, summarize policies, or narrow a large catalog.
That opportunity depends on accessible design across the complete flow. If the payment confirmation, authentication challenge, or error recovery process is inaccessible, the conversational front end merely moves the barrier to a later stage.

---

Enterprise Impact and Competitive Implications​

For enterprise retailers, agentic commerce creates a new distribution channel and a new dependency. Brands gain access to shoppers at the moment they express intent, but they may surrender control over the interface in which products are compared and selected.
This tension resembles the earlier rise of marketplaces and social commerce. Third-party platforms can generate demand quickly, yet they can also weaken direct traffic, compress brand differentiation, and change the economics of customer acquisition.

A new optimization discipline​

Search-engine optimization taught brands to structure pages for crawlers and ranking systems. Marketplace optimization emphasized titles, attributes, reviews, fulfillment performance, and conversion. Agentic commerce will produce its own discipline, sometimes described as AI visibility or agent optimization.
The objective will not simply be to repeat keywords. Brands will need complete, internally consistent product truth that helps an agent answer detailed questions confidently.
Success may depend on factors such as:

• Whether product attributes match the shopper’s constraints.
• Whether inventory and delivery promises are dependable.
• Whether return and support policies reduce perceived risk.
• Whether the merchant has a strong record of successful fulfillment.
• Whether catalog information is fresh and technically accessible.
• Whether the brand permits native checkout for the relevant product and region.

This could benefit retailers with disciplined data operations, even when they lack the advertising budgets of larger rivals. It could also create a new class of optimization services, analytics platforms, and paid-placement strategies.

Competition among commerce intermediaries​

ESW is entering a field that includes ecommerce platforms, payment companies, marketplace operators, cloud providers, and specialist integration vendors. Shopify, PayPal, Stripe, Microsoft, Google, Amazon, and others all have reasons to control or influence the agentic transaction layer.
The strategic prize is not merely a payment-processing fee. The intermediary that observes intent, comparison, selection, and checkout gains valuable insight into how consumers make decisions.
Retailers should therefore examine data rights carefully. They need to know which party can use interaction data, whether customer relationships remain portable, how attribution works, and whether the platform can introduce competing products during the journey.

Direct storefronts will not disappear​

Agentic commerce is more likely to add another layer to retail than eliminate websites and apps. Shoppers will still visit brand-owned experiences for inspiration, loyalty benefits, detailed configuration, community content, service, and confidence in high-value purchases.
The balance may differ by category. Commodity replenishment and clearly specified products are well suited to agent assistance, while fashion, luxury goods, furniture, and enthusiast hardware often involve exploration that is difficult to compress into a short conversation.

---

Strengths and Opportunities​

ESW’s launch combines a practical initial integration with a broader architectural strategy. Its strongest opportunities come from reducing retailer complexity while expanding the number of places where a brand can complete a sale.

Advantages for participating brands​

• The solution extends existing ecommerce investments rather than demanding wholesale replacement. That can reduce implementation risk and preserve established order, inventory, and fulfillment workflows.
• Embedded checkout can reduce abandonment between recommendation and purchase. Every redirect and repeated form introduces another opportunity for the shopper to leave.
• ESW’s cross-border capabilities can help brands serve international demand. Payments, duties, localization, shipping, compliance, and returns are difficult for a general AI platform to coordinate alone.
• The Agentic Hub is intended to support more than one assistant. A multi-agent architecture could protect retailers from relying entirely on a single consumer platform.
• Conversational discovery can surface products that shoppers struggle to find through filters. Rich intent gives merchants another route to match specialized products with specific needs.
• Better product data can improve channels beyond AI. Catalog remediation may also benefit websites, marketplaces, customer-service systems, advertising feeds, and internal analytics.

Opportunities for Microsoft​

Microsoft gains more transaction-ready inventory for Copilot and another partner capable of handling enterprise complexity. A broader merchant base can make Copilot more useful while helping Microsoft compete with other AI platforms seeking to own the shopping conversation.
The arrangement also supports Microsoft’s argument that agents should connect to existing merchant systems rather than replace them. Retailers are more likely to participate if they retain control over fulfillment, policies, customer relationships, and core operational data.

---

Risks and Concerns​

Agentic commerce compresses a complex decision into a friendly interface, which can make errors harder for users to detect. The industry must address technical reliability, consumer protection, platform power, and security before autonomous purchasing becomes routine.

Material risks for retailers and shoppers​

• Incorrect recommendations can create real financial harm. Compatibility errors, omitted restrictions, and misunderstood requirements become more serious when the agent can immediately transact.
• Catalog inaccuracies can spread at machine speed. A faulty attribute may influence thousands of recommendations before a merchant identifies the problem.
• Embedded checkout can obscure the merchant relationship. Shoppers must know who sold the product, who charged them, and who is responsible for refunds or support.
• Platform ranking systems may disadvantage smaller brands. Merchants could become dependent on opaque eligibility and recommendation rules.
• Fraudsters may target agent workflows. Account takeover, manipulated prompts, poisoned product data, fake merchant records, and automated refund abuse are plausible attack paths.
• Privacy boundaries may become difficult to understand. A shopping conversation can reveal health concerns, household details, financial limitations, travel plans, and personal preferences.
• Cross-border mistakes can trigger compliance failures. Restricted goods, incorrect duties, tax errors, and misleading delivery promises can expose merchants to regulatory and reputational consequences.
• Over-automation can weaken deliberate purchasing. An interface optimized for frictionless conversion may encourage impulse buying or reduce the moment in which users reconsider a purchase.

Accountability cannot be conversational​

When something goes wrong, the shopper needs more than an apology generated by an AI model. The system must preserve logs showing which data was supplied, which product was selected, what the user authorized, how the final price was calculated, and which service executed each step.
Enterprises should demand audit trails that support chargeback investigation, regulatory inquiries, customer complaints, and internal incident response. Agent actions need identities, permissions, timestamps, and transaction-level records just like actions performed by human employees or conventional software services.

---

What to Watch Next​

The U.S. launch is only the first phase. ESW says broader availability and additional agent integrations are planned, making geographic expansion and platform diversification the clearest indicators of whether the Agentic Hub can deliver on its design.

Signals that will determine commercial traction​

The first metric to watch is retailer adoption. A platform can announce technical availability, but the channel only becomes meaningful when recognizable brands expose substantial catalogs, enable native transactions, and continue participating after initial tests.
Conversion and return rates will matter as much as transaction volume. If agent-recommended purchases convert well but generate unusually high returns, the system may be matching products too aggressively or failing to communicate fit, compatibility, and policy details.
Microsoft and ESW will also need to demonstrate how merchandising works. Retailers will want reporting that explains which queries surfaced their products, why recommendations succeeded or failed, how much revenue was incremental, and how attribution compares with conventional search or advertising.

Expansion beyond the United States​

International availability will be a defining test because cross-border complexity is central to ESW’s value proposition. Moving into additional markets will require localized consumer experiences, regional payment methods, regulatory alignment, accurate landed-cost calculations, and dependable delivery estimates.
The rollout may proceed unevenly. Different jurisdictions impose different requirements around consumer rights, privacy, automated decision-making, payment authentication, product safety, and disclosures.

The arrival of more agents​

ESW has not framed Copilot as the final destination. Integrations with additional AI platforms would validate the Agentic Hub’s claim to be a reusable infrastructure layer rather than a single-channel connector.
Retailers should watch whether those integrations share a common catalog and transaction model or require substantial platform-specific work. Genuine interoperability would lower costs, while fragmented implementations could recreate the channel-management burden that the hub is supposed to solve.

Standards, controls, and Windows integration​

Protocol support will become increasingly important as AI agents learn to communicate with merchant tools and payment systems. Retailers will favor standards that provide scoped authorization, clear consent, secure identity, revocation, and portable transaction records.
Windows users should also watch how deeply shopping appears across Copilot surfaces. Microsoft must strike a careful balance: commerce features can make Copilot more useful, but aggressive recommendations or poorly labeled promotions could make the assistant feel like an advertising channel embedded in everyday computing.
ESW Agentic Commerce is an early but consequential attempt to build the transactional plumbing behind AI shopping, pairing Microsoft Copilot’s conversational reach with infrastructure for catalog integration, checkout, payments, and international operations. Its success will depend less on the novelty of buying inside a chat than on whether every hidden system can produce accurate products, transparent recommendations, secure authorization, reliable fulfillment, and accountable support. If ESW and Microsoft can meet those requirements while preserving retailer control and consumer trust, agentic commerce could become a durable new route to market; if they cannot, the industry may discover that removing clicks is far easier than removing complexity.

---

Update: Additional details (July 21, 2026)​

AI Journal reports that ESW is wholly owned by Asendia, the international ecommerce logistics joint venture formed by La Poste and Swiss Post. ESW plans to connect additional AI-agent platforms to its Agentic Hub “over the coming months,” although it has not named those platforms or provided a more precise rollout schedule. The hub will also apply defined policies, monitoring, controls, and escalation procedures to automated commerce activity.
Microsoft’s existing Copilot Checkout ecosystem includes PayPal, Shopify, and Stripe, while Microsoft Merchant Center supplies structured merchant data and commerce settings. Microsoft reportedly does not currently impose commissions or affiliate fees on Copilot Checkout transactions. ESW can additionally provide Merchant of Record services, assuming specified responsibilities for payments, taxation, duties, regulatory compliance, and financial liability in supported markets.

ESW launches agentic commerce with Microsoft Copilot, connecting AI-driven discovery to checkout | The AI Journal

New ESW solution launches initially with Microsoft Copilot, enabling brands to make their products discoverable and complete transactions in AI-powered

aijourn.com

---

References​

1. Primary source: 富途牛牛
   Published: 2026-07-21T13:55:00+00:00
   
   Document
2. Related coverage: mckinsey.com
   
   https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-agentic-commerce-opportunity-how-ai-agents-are-ushering-in-a-new-era-for-consumers-and-merchants
3. Official source: news.microsoft.com
   
   

   Microsoft propels retail forward with agentic AI capabilities that power intelligent automation for every retail function - Source

   news.microsoft.com
4. Official source: about.ads.microsoft.com
   
   

   Agentic Commerce | Microsoft Advertising

   Get started with UCP setup Microsoft Merchant Center and enable Copilot Checkout either through MMC or a trusted partner.

   about.ads.microsoft.com
5. Related coverage: mckinsey.com.br
   
   https://www.mckinsey.com.br/our-insights/the-automation-curve-in-agentic-commerce
6. Official source: learn.microsoft.com
   
   

   Dynamics 365 Commerce Model Context Protocol (MCP) server (preview) - Commerce | Dynamics 365 | Microsoft Learn

   Learn how the Dynamics 365 Commerce MCP server (preview) enables agentic commerce experiences for retailers, including tools, authenticated shopper workflows, client integration, and self-service enablement on a Commerce Scale Unit.

   learn.microsoft.com

 

[ChatGPT]

Story update: Additional details — the article above has been updated.