Market insights co-authored by Curator.to and UpScaleX
TL;DR
Agentic commerce represents a fundamental shift from search-based shopping to intent-driven execution. Instead of navigating websites, consumers will increasingly delegate purchasing decisions to AI agents that understand context, compare options, and complete transactions autonomously.
This reshapes retail competition. Data quality, API reliability, and machine-readable trust signals become the new differentiators. Agents need infrastructure they can act on with confidence. Protocols like ACP (Agentic Commerce Protocol) and TAP (Trusted Agent Protocol) serve as foundational infrastructure that enables agents to plan, compare, and transact across merchants.
Over the next two years, retailers will restructure around these systems. The winners will be those who prioritize transparency, interoperability, and execution reliability over traditional marketing tactics.
Consider a shopping experience that starts with a clear instruction instead of a blank search box:
"Get my two kids ready for the school year—clothes, shoes, backpacks, supplies—keep it under $600, prioritize durable basics over trends, and stick to brands that usually fit them."
From that moment, the work shifts from person to system. An agent pulls from the marketplace catalog, infers sizes from past purchases and return history, compares prices across merchants, applies loyalty credits and coupons, and sequences deliveries so essentials arrive before the first day of school. It makes trade-offs automatically: paying a bit more for shoes that historically last longer, saving on items the kids are likely to outgrow, and surfacing only a small set of options for you to approve.
This is agentic commerce: outcomes expressed in natural language, executed end-to-end across catalogs, merchants, and services with minimal human micromanagement. This scenario isn't hypothetical: companies like Perplexity Shopping, Shopify's Sidekick, and Instacart's AI assistant are already enabling versions of this workflow today.
The momentum is undeniable: Bain reported that the number of shopping related queries doubled the past six months and 30-45% of consumers in the United States use generative AI tools like ChatGPT for product research and comparison today.
What is Agentic Commerce
We've already seen the shift from physical to digital storefronts unlock unprecedented distribution and access. In traditional commerce, shoppers relied on in-person assistance and personal trust, whereas e-commerce introduced global reach but often at the cost of personalization. Now we're entering a new frontier of commerce, where intelligence does the same thing for decision-making: AI agents shorten the path to purchase by reducing the cognitive effort required, increasing consumer confidence, and ultimately lowering cost aversion. This is more than an incremental step in retail tech; it's a fundamental rethinking of shopping itself into an integrated, intent-driven flow spanning discovery through checkout.
Why Now:
From an investor's perspective, the rise of agentic commerce represents more than an evolution of the shopping experience; it signals a foundational change in how retail markets function. For the first twenty years of e-commerce, competitive advantage was driven by traffic, SEO, and funnel optimization. As consumers increasingly allow AI systems to interpret intent, compare alternatives, and execute transactions, the sources of advantage shift decisively toward machine-readable infrastructure. Retailers that previously depended on branded content and merchandising must now compete through data freshness, API reliability, identity assurances, and transparent business logic that agents can trust.
This creates a clear directional bet. The inflection point in retail will not be driven by front-end interfaces, but by the emergence of interoperable standards that allow agents to act with confidence across heterogeneous merchant systems. Companies that view themselves not only as storefronts but as machine-facing services will be positioned to capture disproportionately large gains as agentic workflows become the default mode of shopping.
Key Pain Points in Retail Today
Fragmented Customer Journeys & Discovery Friction
Today's online shopping is often fragmented and time-consuming – consumers bounce between search engines, retailer sites, and review pages to find what they need. This cumbersome process leads to drop-offs and lost sales. An autonomous agent streamlines discovery by aggregating information and presenting the best options in one conversational interaction. For example, instead of a shopper manually comparing dozens of product pages, an AI agent can instantly sift through listings, read reviews, and return a succinct, personalized recommendation.
The McKinsey AI Discovery Survey found 44% of users who have tried AI search now prefer it as their primary way to find information, indicating that human shoppers appreciate having an intelligent "guide" cut through the noise. Agentic commerce takes this further by not only guiding discovery but completing the loop (from product search to purchase) in a unified flow, solving the discovery and research pain point.
Erosion of Brand Loyalty and Trust
With endless choices online, retailers have found it challenging to maintain loyalty as many shoppers default to price or convenience over brand relationships. Agentic commerce has a nuanced impact here. On one hand, if consumers delegate decisions to objective third-party agents, brand differentiation might diminish (since the agent could choose any brand that fits the criteria). This is a legitimate concern: Bain's research notes third-party AI agents threaten to commoditize retailers, potentially relegating some to mere fulfillment providers. On the other hand, agentic commerce rewards true customer-centricity. If a retailer can earn an agent's trust by providing superior data (accurate inventory, transparent pricing, APIs for loyalty perks), that retailer's offerings will be favored by the algorithms. In effect, brands must prove their trustworthiness and value to algorithms as much as to humans. The pain point today is that trust signals (reviews, SEO content) are often manipulated or superficial. Agentic commerce pushes retailers to address this by building systems that are transparent and reliable enough for AI agents to consistently recommend them. Those who do will strengthen loyalty in an AI-mediated world; those who don't risk being filtered out. In short, agentic commerce can either erode loyalty or enhance it, depending on how proactively a retailer responds – making it a critical strategic pain point to tackle now.
Lack of Personalization & Generic Experiences
In the e-commerce era, many retailers struggle to recreate the personal touch of the in-store experience. Websites present largely static catalogs and one-size-fits-all interfaces. Agentic commerce offers a remedy: AI assistants that know individual shoppers' preferences, context, and history can deliver truly personalized shopping experiences at scale. Instead of relying on the customer to use filters or search keywords, the agent can interpret a natural language request (e.g. "find a birthday gift under $100 for my friend who loves eco-friendly products") and cross-match it with the shopper's past behavior to find the perfect item. This level of personalization, historically available only from a great sales associate or through painstaking data science, can now be automated by AI.
Emerging Solutions
Pain points above aren't unsolvable; in fact, the infrastructure to address them is already emerging. Here's how leading players are building the rails for agentic commerce:
Data Feeds designed for agents
On the retailer side, enabling agentic commerce isn't just about letting ChatGPT scrape your website, it means providing machine-readable, real-time data and logic through metadata or API endpoints that AI agents can query. Agents struggle to navigate CAPTCHAs and complex 2FA logins designed to stop bots, creating a paradoxical environment where helpful agents are blocked by security measures meant to stop malicious ones. Traditional e-commerce content built on HTML with marketing copy is not sufficient whereas agents need structured data (like attributes, availability, reviews, etc.) without parsing human-oriented text. Trusted sources now carry a higher premium that will impact marketing models through an environment where visibility is earned through API reliability rather than keyword density.
Personalization Engines
In the agentic era, personalization evolves from a marketing tactic to a core operating capability. Today's retailers are often limited by fragmented data silos and basic recommendation algorithms. Agentic AI shatters these limitations by leveraging persistent memory—retaining a holistic history of user preferences and context that informs every decision. This shift enables continuous, real-time learning, allowing brands to move from broad segmentation to hyper-adaptive, individual execution.
In store point of service
Agentic commerce dissolves the boundary between digital intent and physical fulfillment, rendering the traditional, siloed Point of Service (POS) obsolete. Currently, brick-and-mortar environments act as "context black holes" where rich digital histories vanish upon entry. By integrating persistent customer context into the physical infrastructure, retailers transform the store from a passive destination into a responsive, digitally-aware fulfillment node that matches the algorithmic precision of e-commerce.
The Infra of Autonomy: Leading Players & Protocols
Agent Commerce Protocol
OpenAI introduced the Agentic Commerce Protocol (ACP) as an open standard for instant checkout in ChatGPT. ACP defines how an AI agent can move from conversation to transaction, handling product selection and payment in one seamless flow. The real power of these standards is interoperability: rather than custom-coding an integration for every retailer or payment gateway, an agent can use a standard protocol to interact with any system that implements it. PayPal's adoption of ACP is a case in point. By implementing the protocol on their side, PayPal provides ChatGPT one unified interface to access millions of merchants' product catalogs and process payments securely.
ChatGPT Apps SDK
OpenAI's recently released ChatGPT Apps SDK is built on MCP, enabling developers to create ChatGPT plugins (now called "apps") that interface with services like shopping catalogs, booking systems, maps, etc., through natural language. This SDK fundamentally changes how businesses occupy real estate inside an LLM by introducing UI widgets which allow developers to inject interactive HTML/Javascript components directly into the chat stream.
Trusted Agent Protocol (TAP)
Visa attacks the "Trust Gap" with the Trusted Agent Protocol (TAP). Throughout 2024, merchants aggressively blocked AI agents, unable to distinguish between helpful shoppers and malicious scrapers. TAP functions as a digital passport for AI, verifying the identity of the agent—confirming, for instance, that a bot is a verified shopper from Perplexity acting for a specific user—and passing this cryptographically signed verification to the merchant's fraud filters. This allows retailers to whitelist "high-intent" agents while maintaining defenses against bot farms.
What Happens Next:
Looking ahead, expect the next two years to bring several highly reliable shifts that will reshape both consumer behavior and merchant strategy. The first shift is the expansion of AI from recommendation to autonomous execution. Today's systems excel primarily at search, comparison, and summarization. With rapid improvements in long-horizon planning and persistent memory, agents will begin managing entire shopping workflows, from budget allocation to cross-category bundling to dynamic decision-making based on real-world constraints such as delivery timing and availability. The consumer relationship with shopping will increasingly resemble a delegation model rather than an interaction model.
The second shift will be the redefinition of retailer differentiation. Once agents mediate most of the journey, traditional brand-building techniques lose influence. Instead of competing for human attention, merchants will compete for the trust of autonomous systems. Retailers with accurate inventory data, consistent APIs, clear loyalty mechanisms, and transparent rules for pricing and fulfillment will naturally be ranked higher by agents seeking reliable outcomes. Data quality, rather than advertising spend, becomes the key driver of visibility.
The third shift will likely occur at the protocol layer, where platforms and payment networks compete to define the standards that agents use to transact. With ACP formalizing execution flows and TAP establishing machine identity, a multi-layer commercial OS is emerging. Major commerce platforms will accelerate the development of agent-native merchant suites, giving businesses standardized ways to expose catalogs, promotions, availability, and fulfillment logic to AI systems. As these standards mature, cross-agent collaboration will become routine, enabling agents to coordinate discovery, purchase, and post-purchase flows across multiple platforms without human intervention. Together, these shifts will compound, creating momentum faster than traditional retail cycles. The infrastructure being built today, from data standards to APIs to protocol adoption, will determine competitive positioning for the next decade.