Agentic AI has set in motion one of the most significant shifts in the digital economy since the dawn of the dot-com era. And while some predict a fully autonomous future, the more compelling opportunity is the one taking shape right now. The agentic shift underway isn’t about replacing how people shop; it’s about evolving and enhancing the experience in ways that unlock new value for consumers, retailers, and brands.
What excites me is not a distant vision of AI acting on its own, but the meaningful, assistive progress already underway. The next decade will be defined by how effectively brands, retailers, and Large Language Model (LLM) platforms apply these capabilities to create trusted, conversational, and data-driven shopping experiences. At Criteo, we believe the real momentum lies in pragmatic innovation that’s grounded in trust, interoperability, and high-quality commerce data. With that in mind, I’d like to share five reasons I’m excited about the future of agentic commerce in 2026 and beyond.
1. Agentic commerce is emerging as an incremental channel, expanding the ecosystem rather than replacing how people shop today.
Having seen multiple technology cycles come and go, I’ve learned to approach the AI conversation with grounded optimism. The past 30 years of digital evolution have shown a consistent pattern: new channels expand the ecosystem and are additive, rarely replacing what came before. Agentic commerce will follow the same trajectory, adding a powerful new layer of engagement, but not a wholesale replacement for existing ones.
Consider retail’s last major transformation. When ecommerce emerged, many predicted the demise of brick-and-mortar. Yet nearly three decades later, physical retail remains dominant, with ecommerce still accounting for just 20% of total retail sales globally, according to eMarketer. Even if it’s projected to rise further in the next five years, Ark Invest’s prediction that 25% of ecommerce will be agentically driven by 2030 would still only translate to slightly above 5% of total retail sales. Similar incremental growth patterns show that mobile didn’t eliminate desktop browsing, nor social commerce displace traditional ecommerce, but rather, each added a new surface area for engagement rather than consolidating it.
The same will hold true for agentic commerce. Predictions of a fully autonomous shopping ecosystem, where AI agents independently manage purchases, overstate how quickly consumers will adopt such behavior. Evidence continues to point to a more gradual, assistive evolution: one where agentic systems enhance research, comparison, and checkout efficiency, but decision-making remains primarily human-led.
Consumers gravitate toward agentic tools when they reduce friction in the shopping journey, such as saving time, simplifying choices, and helping to identify the right products at the right price. By removing barriers, agents can unlock purchases that shoppers may have abandoned previously, effectively expanding overall online shopping rather than shifting it from one channel or another.
These are meaningful gains, but they do not yet signal a shift toward full automation. Today’s LLM platforms are strongest when assisting with research and comparison, while shoppers still want to maintain control over final decisions. As agentic experiences continue to evolve, their uptake will grow in step with how effectively they address these foundational needs.
2. Agentic assistants will introduce a new discovery layer that fragments search even further, making product discoverability the next competitive battleground.
Search has been fragmenting for years, as consumers have shifted away from traditional blue-link Google search toward environments like Reddit for topic-specific, conversational forums with real users. LLM platforms now accelerate this shift, with prompt-based queries functioning as a new layer of search, redistributing discovery across a wider set of touchpoints.
This shift is evident in emerging behavior. A recent Criteo survey of 10,000 respondents found that 40% of U.S. shoppers now use agentic shopping assistants regularly for product research, yet 96% of them also use other channels along the way, including search engines, social platforms, and brand and retailer sites. This is almost identical to the multichannel behavior we saw last year, underscoring that no displacement is occurring. Instead, discovery is becoming increasingly distributed as consumers move across multiple endpoints.
As this fragmentation accelerates, distributed discoverability becomes the new imperative. Products must be findable wherever consumers initiate queries: across retailer front ends, social platforms, LLM platforms, and other emerging AI-driven environments. As part of this shift, we’re leaning into helping our clients’ products become more discoverable across an increasingly fragmented and agentic ecosystem.
3. Agentic experiences will push retailers to modernize their AI-driven front ends, elevating retail media in the process.
There’s a misconception that LLM platforms will cannibalize retail media networks. In practice, these systems will expand the funnel by introducing new entry points for discovery and intent, while retailer environments remain the core of conversion, fulfillment, and loyalty.
As Forrester’s Emily Pfeiffer points out, consumers aren’t looking for a single, generic shopping experience. They now expect personalized, guided discovery, and increasingly, conversational interfaces that mirror the intuitive, context-aware interactions they’re becoming accustomed to on LLM platforms. This raises the UX benchmark for retailers and accelerates the need to modernize front ends across websites, apps, and in-store experiences.
Early traction from retail-owned assistants highlights how they can make online shopping grow faster: during Black Friday and Cyber Monday, Rufus-assisted Amazon sessions resulting in a purchase grew more than 100%, compared with 20% growth for non-Rufus sessions, according to Sensor Tower. Walmart’s Sparky assistant is showing growing traction too. Furthermore, according to Accenture, U.S. shoppers prefer using retailer or brand-specific chat assistants, as opposed to third-party LLM platforms while shopping.
At the same time, a new incremental opportunity is emerging: sponsored recommendations within retailer-owned chatbots, as well as within retailer app integrations on LLM platforms like ChatGPT, where retailers maintain control over product ranking.
The path forward lies in balanced ownership and collaboration. Retailers that build agentic experiences into their own environments, while selectively sharing structured commerce data with LLM platforms, will maintain influence over customer journeys and extend discoverability across an increasingly distributed landscape.
4. Advertising will become the dominant monetization model for LLM platforms, creating a scalable revenue layer.
Across the ecosystem, one question looms large: how will LLM platforms monetize at scale? The most sustainable and flexible model will be advertising, and we’re already seeing that play out today as Google has begun showing ads in its AI Mode search engine. Advertising allows platforms to capture value across the discovery and decision phases of commerce, without needing to own checkout or fulfillment.
As Eric Seufert noted in Mobile Dev Memo, affiliate and marketplace models depend on capturing full transactions, limiting scale and interoperability. Advertising, by contrast, monetizes attention and intent. Contextual placements within conversational experiences create recurring, high-margin revenue opportunities that can grow alongside user engagement. And when native ads are highly relevant and contextually aligned, they integrate so seamlessly into the experience that they feel more like useful content than disruption, making them a powerful and scalable monetization lever.
ChatGPT exemplifies this dynamic. Only about 5 % of its users pay for subscriptions, leaving the vast majority in a free tier where ads will be the logical engine for growth. Advertising remains the most proven and scalable mechanism for digital monetization, and in the GenAI era, it will reward transparency, data quality, and relevance.
5. The real competitive advantage in agentic commerce will come from quality commerce data.
I recently had a firsthand reminder of the gaps that still exist in AI-powered product discovery. As a cyclist in New York City, I spent nearly an hour in an LLM platform searching for “puncture-proof, durable city-bike tires.” The reasoning was impressive, but the shopping experience wasn’t. The assistant surfaced broken links, discontinued models, and incomplete product information. In the end, I went to my local bike shop, where I trust the people and the process.
This experience underscores a broader truth about the state of AI today: while intelligent assistants excel at conversation, their recommendations falter without access to high-quality, structured, real-time commerce data. And when those recommendations fall short, they don’t just miss a sale; they erode consumer trust. Recent analysis from OpenAI shows that ChatGPT shopping research delivers only 64% accuracy, illustrating how far these systems remain from reliably supporting end-to-end purchase journeys.
The current state of agentic commerce is that the infrastructure and interoperability needed to deliver relevant product recommendations remain underdeveloped. Most LLM platforms still lack access to real-time inventory, accurate pricing, detailed product attributes, and unified checkout and fulfillment systems. These gaps make high-quality, structured commerce data at scale a critical dependency for AI-driven shopping.
This is also why LLM platforms struggle with conversion today. They can initiate interest, but without real-time commerce data and the optimization systems retailers rely on, they can’t convert interest into completed purchases. Until they close that gap, they will remain powerful demand engines but incomplete commerce platforms.
At Criteo, we have deep visibility into how people buy, informed by 720 million daily active users, interacting across 4.5 billion SKUs in our Universal Product Catalog, and generating more than $1 trillion in ecommerce transactions each year. This scale creates a commerce-intelligence foundation that is continually refreshed. By maintaining consistent, detailed metadata across global retail feeds, it enables systems to surface only products that are in stock, accurately described, and verified by actual buyers. This level of precision minimizes friction in discovery, improves recommendation accuracy and builds the long-term trust required to convert and retain consumers.
The path forward for agentic commerce
The next phase of agentic commerce will be shaped by several converging dynamics: increasingly distributed discovery as search fragments across channels, the emergence of AI-driven retail experiences, advertising solidifying as the scalable monetization model for LLM platforms, and the growing importance of high-quality commerce data to ensure accuracy and trust.
At Criteo, agentic AI already informs how our own AI agents collaborate and how intelligence flows across our platform, driven by our early investment in our Model Context Protocol (MCP), which enables interoperability across systems. Together, these shifts point to an ecosystem where assistive intelligence plays a central and effective role in how people discover, evaluate, and choose products.
Criteo’s roadmap embraces this evolution and prioritizes scalable mechanisms that strengthen discoverability across the ecosystem, from sponsored product placements in retailer-owned assistants and AI-integrated retail apps to AI shopping infrastructure and commerce data feeds.
Criteo’s proven cross-channel model uniquely positions us to help clients capitalize on the incremental opportunities created by agentic AI. The future of agentic commerce will be led by those that deliver accurate, trustworthy, and interoperable pathways to product discovery. That is the foundation Criteo is building, and the intelligence layer that will support the next decade of commerce.
