Agentic commerce: How AI agents are reshaping the future of shopping

The shopping experience has always evolved with the advent of new technology, and each advance has changed what consumers expect from brands. Agentic commerce is the latest evolution: AI agents that handle the research, comparison, and purchase autonomously, often without the consumer ever visiting a brand's site.

As AI becomes a primary interface for customers. Maintaining the engagement and personalization layer with customers keeps the brand relationship intact, however, the transaction happens.

TL;DR

• Agentic commerce is the use of autonomous AI agents to research, compare, and complete purchases on behalf of consumers or businesses, with minimal human input.
• Consumer adoption of agentic shopping is expected to jump from 19% to 46% by the end of 2026, according to the Braze Retail Customer Engagement Review.
• Agent selection is driven by data quality: machine-readable product feeds, structured attributes, and generative engine optimization determine whether a product appears in an agentic search at all.
• Consumer trust in agents remains cautious. Only 10% of consumers are willing to let agents operate fully independently, meaning most still want oversight of purchasing decisions.
• Brands need to prepare on two fronts: structured data and GEO for agent discoverability, and direct customer relationships for the engagement layer that persists beyond the transaction.

Key takeaways

• 71% of marketing leaders already say AI agents have weakened their ability to connect directly with customers, according to the Braze Retail Customer Engagement Review.
• The payments infrastructure for agentic commerce is live: Google's AP2, Mastercard's Agent Pay, Stripe's temporary virtual cards, and ChatGPT's instant checkout are all operational.
• Brands whose product data is fragmented or incomplete are already losing visibility in agentic searches. Data readiness is a commercial priority, not a technical one.
• The same agent that selects a brand can switch to a competitor just as easily. For subscription and repeat-purchase businesses in particular, the stakes are high.
• The engagement layer, personalization, cross-channel presence, and direct customer trust, carries increasing commercial weight as AI intermediation grows.

What is agentic commerce?

Agentic commerce is projected to generate between $3 trillion and $5 trillion globally, by 2030, according to McKinsey research. As adoption accelerates, brands need a clear understanding of what it is and how it works. For example:

• Nearly half of all consumers are expected to use AI agents for brand interactions by the end of 2026, according to the Braze Retail Customer Engagement Review.
• Only 10% are willing to let those agents operate without oversight.
• 71% of marketing leaders say agents have already weakened their ability to connect directly with customers.

Agentic commerce definition

Agentic commerce is an approach to buying and selling where AI agents autonomously research, compare, negotiate, and complete purchases on behalf of consumers or businesses, usually with minimal human intervention. Unlike traditional eCommerce, which requires manual browsing and checkout, agentic commerce delegates decision-making to intelligent AI systems that reason and act across multiple platforms.

How agentic commerce differs from traditional commerce

Agentic commerce differs from everything that came before it: agents make decisions autonomously compared to customers making purchase decisions independently.

Traditional eCommerce puts every decision on the customer: search, compare, click, add to cart, and pay. But now, 45% of consumers already use AI for some part of the buying journey, according to the IBM Institute for Business Value. Most of that is still assistive: people using AI to research or compare options, not complete the transaction.

Take something as simple as reordering a household staple. A recommendation engine might remind the customer it's running low. A chatbot might help their customer find it. Getting it ordered still falls to the customer. And neither the chatbot nor the customer could adapt if the usual brand was out of stock, the price had jumped, or a better option existed elsewhere.

The main differences come down to three things:

• Autonomy: Agents act within guardrails without waiting for approval every step.
• Reasoning: Agents adapt in real time. If a price drops, a product sells out, or a delivery window closes, the agent adjusts.
• Interoperability: The agents can connect across platforms via APIs to complete full commerce journeys, not just individual steps.

How agentic commerce and AI-powered shopping works

Agentic commerce works by moving a transaction through a connected sequence of stages, from a customer stating a goal or asking a question, all the way through to payment, fulfillment, and post-purchase support. At each stage, the agent handles what would otherwise require manual input, running on three building blocks:

1. Large language models (LLMs) for reasoning;
2. APIs connecting to retailer backends; and
3. Structured product data the agent can read and act on

Customer-to-agent engagement

Customers ask a question or state a need and the agent interprets it. A request like “find me a wireless speaker under $100 with next-day delivery” gives the agent its goal. The buying journey begins with a conversation, rather than a search engine query. If the request is too broad, the agent asks for more information. It also draws on stored preferences, past purchases, and remembered sizes, so the same prompt gets smarter results over time.

Autonomous execution

Once a goal is set, the agent moves through a multi-step workflow, scanning retailers, comparing prices in real-time, checking inventory, applying available discounts, and completing the transaction.

Agent autonomy can be designed via a tiered structure. Routine or low-value purchases can run fully automated. Higher-value transactions might require the person to approve before the final step. The agent narrows down the options, and then the human confirms.

Product discovery and decision-making

Agents query structured data across multiple sources simultaneously, evaluating product attributes, pricing, availability, reviews, and fulfillment options in parallel.

Visibility depends on data quality. This is where generative engine optimization (GEO) can improve the chances of your product or service being seen.

Machine-readable product data, standardized attributes, and clear metadata determine whether a product appears in an agentic search. If product data is incomplete or missing key attributes, the agent either fails the query or deprioritizes that merchant for future searches.

Merchant-to-agent interaction and agent-to-agent commerce

For agents to transact at scale, merchants need machine-readable interfaces: APIs for product catalogs, real-time pricing, inventory levels, return policies, and fulfillment options.

Beyond merchant-to-agent, agent-to-agent commerce is also taking shape, where a consumer's AI agent communicates directly with a merchant's AI agent, negotiating and transacting with no human interface in between. For this to work at scale, agents from different platforms need agreed standards to communicate and transact. Several are already in place:

• MCP- Model Context Protocol (Anthropic): connects agents to data systems including product catalogs and inventory
• ACP (OpenAI and Stripe): standardizes how agents handle payments and checkout
• A2A (Google): covers direct communication between independent AI agents

These standards make multi-agent orchestration possible at scale. One agent represents the buyer and another represents the seller, all operating through a shared protocol.

Agentic payments

Agentic payments work similarly to giving a customer a pre-loaded card with spending rules attached and the APP (agentic payments protocol) infrastructure that makes this possible is already live.

OpenAI and Stripe's Agentic Commerce Protocol (ACP) powers ChatGPT's instant checkout, allowing purchases to be completed directly within a chat interface without the user ever leaving the conversation.

Google's AP2 verifies that an agent is genuinely authorised before a transaction goes through. Mastercard's Agent Pay does the same across its global network. Stripe also generates a temporary card number for each agent transaction so the user's actual payment details are never passed to a retailer.

Every transaction creates a record of what was bought, by which agent, and under what authorization. The payments industry is also developing ways to verify agents themselves, not just the humans behind them.

Post-purchase support

Agents track shipments, manage returns, and handle retailer communications when something goes wrong. A replenishment agent can monitor usage and reorder automatically, within parameters the user has set.

Agents also identify follow-on purchases based on what was bought and when, making recommendations without the person needing to search again.

Agentic commerce use cases across industries

The mechanics of agentic commerce are consistent across industries. What changes is where the friction is, and which tasks agents are being deployed to handle. Here are some examples of how you can use agentic AI in marketing.

Retail and eCommerce

In retail, agentic commerce is already handling the tasks that create the most friction, like recurring purchases, real-time price comparison, and cross-channel fulfillment coordination.

What agents do:

• Monitor household staples and reorder automatically within set budgets
• Compare prices across multiple retailers in real time before committing to a purchase
• Coordinate online ordering with in-store pickup availability across channels

The brand opportunity: Agents that know a customer's preferences and purchase history can act on them immediately, without requiring the customer to re-engage with a brand's channels.

B2B procurement and supply chain

In B2B, agentic commerce is changing how procurement teams handle vendor relationships, pricing, and supply disruptions. 61% of procurement leaders cite geopolitical and supply risks as their top concerns, according to IBM, and agents are increasingly being deployed to respond to exactly those risks in real time.

What agents do:

• Validate new vendors against compliance and quality standards automatically
• Negotiate volume pricing within pre-approved parameters
• Identify and activate alternative sourcing when supply disruptions occur

The brand opportunity: Agents can respond to supply chain disruptions faster than any human procurement team. Decisions that used to take days to make now take only minutes.

Travel and hospitality

In travel, agents are handling the full booking workflow and the moments when plans fall apart.

What agents do:

• Search and book flights, hotels, and transfers based on traveler constraints and preferences
• Automatically rebook when conditions change: cancellations, delays, or significant price drops
• Process refunds within pre-approved limits without requiring human intervention

The brand opportunity: The rebooking moment is where customer loyalty is built or lost. An agent that handles disruption smoothly, without the traveler needing to call customer support, is a brand experience in itself.

Digital subscriptions

In subscription services, agents monitor usage and optimize plans, or switch providers, on the user's behalf.

What agents do:

• Track usage against plan limits and flag over- or under-utilization
• Recommend plan changes based on actual usage data and current pricing
• Initiate provider switches when a better price-to-performance option is available

The brand opportunity: An agent can switch a customer to a competitor as easily as it can renew them. Retention depends on offering demonstrably better value.

QSR and food delivery

In QSR and food delivery, agents are handling meal planning, loyalty optimization, and recurring orders.

What agents do:

• Plan and schedule recurring orders based on household preferences and dietary requirements
• Apply loyalty points and available offers automatically at the point of order
• Reorder from previous baskets without requiring the customer to rebuild the cart

The brand opportunity: Recurring agent-driven orders create predictable demand, but only for brands whose APIs, loyalty data, and menu information are structured for machine access.

Benefits of agentic commerce for brands and consumers

Agentic commerce reduces friction for consumers while opening new commercial opportunities for brands. Here are the ways in which it can help you.

1. Faster transaction processing and reduced checkout friction

Agents remove the friction of multi-step checkout. The transaction happens at the point of decision, without form-filling or repeated logins.

2. Scalable, real-time personalization of offers

Agents remember preferences, sizes, and past purchases, delivering a personal shopper experience without the manual effort. Every interaction can be tailored to the individual at a scale any human team would struggle to replicate.

3. Reduced search time and improved customer decision-making

Instead of comparing options across multiple sites, consumers receive filtered, reasoned recommendations based on their specific constraints. The agent handles the research; the consumer makes the final call.

4. New product discovery pathways and monetization of agent interactions

Brands can reach consumers through agent ecosystems and LLM-based shopping assistants and interfaces that didn't exist two years ago. These interactions also create new monetization models, from sponsored suggestions to fee-based agent access.

5. Operational accuracy with less manual oversight

Agents are designed to apply rules consistently, significantly minimizing the risk of manualfatigue errors. For businesses running complex procurement or subscription management, this means processes run accurately with less manual oversight.

Challenges and limitations

Agentic commerce creates real opportunities but also introduces complications that brands, retailers, and payment providers are still working through.

Data readiness

If a retailer's product catalog spans multiple systems with inconsistent attributes, incomplete specifications, or outdated pricing, AI agents can't evaluate those products reliably. Fragmented data limits both discoverability and interoperability, and an agent that can't access clean, standardized product information will either return poor results or skip that merchant entirely.

The same problem applies internally. Agents making decisions on behalf of customers need access to rich, unified data to do it well, like purchase history, preferences, behavioral signals, and loyalty status. If that data sits in disconnected systems, the agent is working with an incomplete picture, and the decisions it makes will reflect that.

Consumer trust

Consumer willingness to hand control to AI agents is still cautious. 83% of consumers share concerns about privacy, data misuse, and unsolicited marketing, according to the IBM Institute for Business Value. The Braze Retail Customer Engagement Review also found that only 10% of consumers are willing to let agents operate fully independently, meaning the vast majority still want to stay in the loop for most purchasing decisions.

Brand visibility risk

As agents mediate more transactions, direct touchpoints between brands and customers shrink. The discovery moment, the browsing experience, and the checkout interaction are all points where brands build familiarity and preference. The Braze Retail Customer Engagement Review found that 71% of marketing leaders say agents have already weakened their ability to connect directly with customers.

Legacy infrastructure

Existing fraud detection and payment authentication systems were built around human behavior. The signals they use to verify intent and flag anomalies, like browsing patterns, session duration, and device fingerprinting, don't translate cleanly to machine-initiated transactions. Adapting those frameworks to recognize and trust AI intermediaries is an active challenge across the industry, which is why standards like Google's AP2 and Mastercard's Know Your Agent protocol are being developed alongside the payment infrastructure itself.

How brands can prepare for agentic commerce

Preparing for agentic commerce spans data infrastructure, customer relationships, and how brands think about discoverability. Each area requires a different kind of investment, but they all point in the same direction.

Standardize product data

Product data needs to work for both human browsers and AI agents. Complete, structured attributes, accurate real-time inventory, and standardized taxonomy are what allow agents to find, evaluate, and act on a product. Without them, the product is invisible to an agentic search.

Integrate open APIs

Agents transact through programmatic interfaces, not browser sessions. Brands need well-documented APIs that return real-time pricing, inventory, and fulfillment data. The MCP and ACP protocols define how agents connect to merchant systems, but the underlying data infrastructure needs to exist first.

Invest in direct customer relationships

The 2026 Braze Customer Engagement Review found that 43% of consumers would stop engaging with a brand entirely if their personal data were misused. As agents mediate more transactions, the direct touchpoints where brands build and repair that trust naturally decrease, making the relationships brands invest in now even more commercially significant.

Adopt platforms that connect with LLM ecosystems

The engagement layer becomes more important as AI intermediation grows. Platforms that connect with LLM commerce ecosystems allow brands to stay present and personalized across both direct and agent-mediated commerce journeys. Braze AI agents are built for exactly this, and the Braze integration with ChatGPT is a live example of what brand presence inside an LLM ecosystem looks like.

Rethink discoverability for AI

Generative engine optimization is the practice of structuring content so AI agents can find, evaluate, and act on it. Where traditional SEO optimized for clicks, GEO optimizes for agent selection: machine-readable product feeds, structured metadata, and content that an LLM can cite and act on without visiting the page.

How Braze supports the agentic AI eCommerce era

Braze approaches agentic commerce as an engagement challenge. As transactions move through AI agents, the platform provides the tools to keep brands personalized, present, and in control of the customer relationship.

• BrazeAI Agent Console™: The Agent Console is a centralized environment for building, managing, and deploying AI agents that generate content, interpret data, and adapt campaigns in real-time. Rather than sending static campaigns,custom agents build and send individualized messages,informed by each customer's behavior, purchase history, and preferences.
• Braze SDK integration with ChatGPT: The Braze integration with ChatGPT lets brands build custom apps that operate directly within ChatGPT's interface, extending brand presence into one of the most widely used AI platforms. Brands can create interactive product carousels, storefront details, and personalized experiences inside a conversation, reaching consumers at the point where they're already using AI to research and decide.
• BrazeAI Decisioning Studio™: Decisioning Studio uses reinforcement learning to automatically determine the optimal action for each individual customer, personalizing the channel, message, offer, timing, and frequency based on real-time, first-party data. It acts as the intelligence layer that keeps brand communications relevant and timely even as the transaction itself moves through an intermediary.
• BrazeAI Operator™: Operator is an in-dashboard assistant that takes direct action inside Braze, from building content and writing personalization logic to creating entire customer journeys. It can also create agents that run through the Agent Console so marketing teams can deploy more sophisticated AI capabilities without needing additional technical resources.
• Cross-channel orchestration: As agentic AI in marketing becomes more central to how brands operate, Braze's cross-channel orchestration connects messaging across email, push, SMS, in-app, and emerging channels to maintain consistent brand relationships alongside agent-mediated transactions. Wherever the customer is, and however they arrive there, the engagement layer stays intact.

Final thoughts and takeaways

Agentic commerce is moving the purchase decision from the human browser to the AI agent, and adoption is accelerating faster than most brands have planned for. The Braze Retail Customer Engagement Review found that consumer adoption of agentic shopping is expected to jump from 19% to 46% by the end of 2026.

Brands that unify their data, build machine-readable product feeds, and invest in direct customer relationships now will have a structural advantage as agentic commerce scales.

Two things determine a brand's position in an agentic commerce world. The first is discoverability: Machine-readable product data, structured attributes, and generative engine optimization are what allow agents to find and evaluate a product. The second is the relationship that exists beyond the transaction: The personalization, cross-channel messaging, and direct customer trust that sits outside what agents mediate.

As AI becomes a primary interface for customers. That engagement layer carries increasing commercial weight. The brands investing in it now are building something agents can't replicate and competitors can't easily displace.