Agentic Commerce Is Here: What It Means for Product Discovery in 2026

At Shoptalk Spring 2026, something clicked. Nobody talked about AI as a future technology. Gap, Walmart, Sephora: they all presented AI as infrastructure that’s already live, already measured, already driving revenue. The experimentation phase? Over.

The big theme on everyone’s lips: agentic commerce. AI agents that don’t just recommend products, but browse, compare, and buy on behalf of consumers. Entirely autonomously.

The numbers tell the story. McKinsey forecasts that AI agents will orchestrate $900 billion to $1 trillion in US retail revenue by 2030. Around 45 to 50 million Americans already use AI tools for shopping. And conversions from AI referrals shot up by 1,247% in late 2025.

This isn’t a trend to watch. It’s one to act on. And for retailers who care about product discovery, personalization, and conversion, the implications are massive. At Crobox, we’ve been thinking about this shift for a while, because it touches every part of what we help our clients build.

What Is Agentic Commerce?

Traditional eCommerce puts the shopper in control. They search, filter, compare, read reviews, click “buy.” Guided selling improved on that by helping shoppers navigate complex choices through intelligent recommendations and product finders.

Agentic commerce goes further. The AI agent handles the full journey. A consumer says “I need trail running shoes under $150” and the agent researches options across multiple stores, compares them, and completes the purchase. The consumer barely lifts a finger.

It’s already happening at scale:

• Google unveiled its Universal Commerce Protocol at NRF in January 2026, co-developed with Shopify, Etsy, Target, and Walmart. Its “Buy for Me” feature lets agents complete purchases directly from search results.
• Walmart plugged its catalog into ChatGPT for conversational shopping, from discovery to checkout.
• Sephora connected its loyalty program and product catalog to ChatGPT’s interface.
• Macy’s launched “Ask Macy’s,” a Gemini-powered bot for navigating their catalog through natural conversation.

Amazon, meanwhile, has blocked external AI agents from accessing its site. Probably to protect its $56 billion ad business, which depends on people browsing Amazon directly. But they countered with their own agent that shops other retailers’ sites when Amazon can’t fulfill a request.

The platforms have made their bets. Agents are the next shopping interface.

Your Product Page Might Get Skipped Entirely

Here’s the uncomfortable part. If AI agents become the main way people discover and buy, the product page, where your personalization, nudging, and conversion optimization lives, could be bypassed completely.

Think about it:

1. A consumer tells the agent what they want
2. The agent queries catalogs, reads reviews, compares specs
3. The agent recommends a product and buys it (with approval)

The consumer never visits your site. They never see your product badges, your dynamic messaging, your behavioral nudges. The agent sees raw data: attributes, specs, review sentiment, pricing. And it makes decisions based on that.

The question retailers need to ask is no longer just “How do we convince shoppers?” It’s also “How do we convince agents?”

This is where Crobox clients are already ahead. Our platform enriches product data with customer-centric attributes and behavioral insights, exactly the kind of structured, intent-mapped information that AI agents need to accurately recommend your products. More on that below.

Product Data Is Your New Competitive Moat

When agents do the shopping, your product data becomes the most important asset you have. Not your landing page. Not your ad creative. Your data.

AI agents don’t respond to a pretty hero image or an urgency badge saying “Only 3 left!” They parse structured information: product attributes, feature descriptions, compatibility specs, review sentiment. And they use it to match products against what the consumer actually asked for.

To win in agentic commerce, your product data needs to be:

• Rich and structured. Go beyond title, price, and image. Include detailed attributes, benefits, use cases, and compatibility information that agents can parse and compare.
• Customer-centric, not technical. Your internal spec sheets aren’t how customers describe what they need. When someone tells an agent “I need shoes for long runs on pavement,” the agent has to match that against your data. If your data only says “10mm drop, EVA midsole,” you’ll lose to the competitor whose data says “built for road running, cushioned for long distances.”
• Continuously enriched. Static product data goes stale. Behavioral signals, review sentiment, and purchase patterns should feed back into your catalog to keep it sharp.

This is the core of what Crobox’s Product Data Enrichment does. We use AI combined with real consumer behavior data to transform raw, technical product catalogs into rich, customer-facing attribute sets. Our clients like ASICS use this to power their product finders. But in the agentic commerce world, this enriched data also becomes what makes your products visible and competitive when AI agents are doing the selection.

Humans Still Make the Final Call

Don’t go all-in on agent optimization and forget about actual people. That would be a mistake.

Most shopping decisions are still made by humans. And even in agent-assisted journeys, there are critical moments where the person takes back control. They review what the agent recommended. They compare the final two options. They decide whether to actually buy.

These are the moments where guided selling still wins. When a shopper lands on your site, whether from an agent’s recommendation, Google, or Instagram, the experience needs to:

• Reduce choice overload with intelligent filtering and recommendations
• Build confidence through personalized messaging and social proof
• Guide the decision with product finders that replicate what a great sales associate does in-store
• Minimize friction to reduce cart abandonment

Brands doing this well are seeing conversion rates of 6.8% on average, with the top 10% hitting 14.3%. The human experience still matters. Enormously.

At Crobox, this dual approach is built into how we work. Our Product Finders guide humans through complex purchase decisions, while our data enrichment layer makes sure your products are also optimized for the agents sending those humans your way. You need both.

The EU AI Act: Regulation That Plays in Your Favor

Most retailers aren’t paying enough attention to this: the EU AI Act becomes fully applicable on August 2, 2026. That’s less than four months away.

For eCommerce personalization, it means:

• Subliminal manipulation is banned. AI personalization that targets psychological vulnerabilities in ways consumers can’t perceive or resist is now illegal for EU-facing campaigns.
• Dynamic pricing faces scrutiny. Pricing that discriminates based on consumer vulnerability will trigger compliance requirements.
• Transparency is mandatory. Consumers need to know when they’re interacting with AI and how recommendations are generated.

On top of that, the Digital Fairness Act is expected in Q4 2026, targeting dark patterns, misleading personalization, and addictive design. The EU estimates dark patterns cause 7.9 billion euros per year in consumer harm.

This might sound like a headache. But think of it differently: compliant personalization is a competitive moat.

Companies with mature first-party data programs achieve 2.9x higher revenue growth versus those relying on third-party data. Privacy-first strategies report 43% higher customer lifetime value. The brands that build transparent, consent-driven personalization now will have a huge advantage when enforcement kicks in.

This is where the line between ethical nudging and dark patterns really matters. Ethical nudging helps consumers make better decisions by reducing choice paralysis and surfacing relevant information. Dark patterns exploit cognitive biases to push people toward choices that benefit the retailer at their expense.

At Crobox, we’ve always been on the ethical side of that line. Our behavioral psychology approach is designed to empower shoppers, not manipulate them. We don’t collect PII. We work with anonymized behavioral signals. As regulation tightens, this isn’t just a philosophical stance anymore. It’s a business advantage that our clients benefit from directly.

5 Things You Should Do Right Now

The shift to agentic commerce won’t happen overnight. But the retailers who prepare now will be in a much stronger position. Here’s where to start:

1. Audit and Enrich Your Product Data

Your product data needs to be machine-readable, customer-centric, and rich enough for AI agents to match against consumer intent. Product data enrichment that captures benefits, use cases, and compatibility in the language your customers actually use is no longer optional. Crobox can help you get there fast, with AI-powered enrichment validated by real consumer behavior.

2. Build for Both Agents and Humans

Create a dual-layer strategy. Structured, enriched data that agents can consume. Guided, personalized experiences for the shoppers who land on your site. With Crobox, you get both from a single platform: data enrichment for the agent layer, Product Finders and dynamic messaging for the human layer.

3. Get Ahead of EU AI Act Compliance

August 2026 is right around the corner. Audit your personalization for anything that could be classified as subliminal manipulation or a dark pattern. If your current vendor can’t clearly explain how their personalization works and where the data comes from, that’s a red flag.

4. Invest in Zero-Party Data

With third-party cookies eroding and regulation tightening, zero-party data, the information consumers willingly share, becomes your most reliable data source. Product finders, quizzes, and guided selling experiences are natural collection mechanisms. Every Crobox Product Finder doubles as a zero-party data engine, giving you rich consumer insights while providing shoppers with genuinely helpful guidance.

5. Rethink Your Discovery Metrics

Page views, time on site, and click-through rates become less meaningful when agents handle discovery. Start tracking how often your products get recommended by AI agents, your product data quality scores, and conversion rates from agent-referred traffic.

The Future Is Hybrid

Agentic commerce isn’t the end of product discovery. It’s the start of a more layered version. Retailers need to optimize for two audiences at once: the AI agents orchestrating more and more shopping journeys, and the humans who still make the final call.

The good news? The fundamentals haven’t changed. Understanding your customers, enriching your product data, reducing decision friction, building trust through transparent personalization. These matter just as much for agents as they do for people.

If you want to get ahead of this shift, we’d love to talk. Crobox helps retailers build product discovery experiences that work for both humans and machines, with enriched product data, intelligent guided selling, and privacy-compliant personalization built in from the ground up.

Book a demo and let’s figure out what agentic commerce means for your business.