By Purva Gupta, Co-Founder & CEO, Lily AI
Consumers are no longer typing short keywords into a search bar. They’re expressing intent in full, nuanced sentences: “I need something polished but comfortable for an outdoor event next week,” or “Find me a gift that feels personal but ships fast.”
AI agents don’t just match these requests to products They interpret them, reason through them, and decide which options are genuinely relevant.
That shift from ranking to reasoning is the core transformation behind agentic shopping. And with Shopify and ChatGPT now enabling checkout directly inside the conversation, this shift is accelerating fast. In Lily AI’s latest consumer research, 70% of shoppers say they’re exploring or plan to explore ChatGPT for holiday shopping, up from roughly 40% at the start of the year. That’s a major behavioral change in a very short amount of time.
I recently joined a conversation with Third and Grove and ULE Group to unpack what this partnership means for ecommerce brands and where the biggest opportunities lie. Here are six takeaways every retailer should be thinking about right now.
AI recommends what it understands
Product content used to serve humans first and search engines second. Agentic shopping flips that relationship: your product data now has to serve both, simultaneously and with clarity.
Vague language such as “great for all occasions” no longer helps a machine understand context. AI needs explicit cues (occasions, textures, fits, climates, sentiments, benefits) all mapped to structured attributes that can be parsed instantly.
- Content must be unambiguous
- Attributes must be exhaustive
- Language must reflect how people actually express needs
- Data must be machine-readable, not just customer-friendly
If the product content can’t answer the user’s intent directly, the AI simply moves on (often to a competitor whose data is clearer and more complete).
In an agentic environment, brand also becomes a trust filter. AI can surface dozens of options, but consumers will only act on the ones they recognize and believe in. Clear, consistent product data and a credible brand work together: the data gets you into the answer, the brand closes the loop.
2. Product data can’t be static
Historically, product content has been treated as a periodic task. Something created, published, and revisited only when necessary. But in an agentic environment, product data has to function as a living system.
Consumers change how they describe things week to week. AI models evolve at the same pace. Inventory shifts hourly. Seasonal context moves quickly. And yes, large models expect freshness, sometimes down to minutes, not days.
This is where many brands will struggle. Manually updating content, rewriting descriptions, refreshing feeds, and expanding attributes isn’t sustainable at scale. The volume and velocity required are simply too high.
Continuous content optimization, enriched, restructured, and refreshed in real time, is the new foundation for discovery. Without it, products become invisible, even when they’re perfect for the shopper’s request.
3. Your website is still important
There’s a misconception that as more shopping happens inside AI interfaces, the brand’s ecommerce site will matter less. In reality, your website is more important than ever because it’s your truth set data source for AI. Every variant, size, description, title, and promotion has to be clearly structured and machine-readable so AI systems can pull the right information instantly. That means clean schema, consistent product attributes and highlights, and real-time updates to inventory and pricing.
When your site data is accurate and connected, agentic shopping and checkout should be able to work far more seamlessly because the AI already trusts your source.
In this new world, your website is your AI data engine so be sure to feed it well. That also extends beyond PDPs: policy pages, FAQs, and reviews all contribute to the trust signals AI agents evaluate when deciding which products to surface.
4. SEO is the foundation, AEO is the evolution
What you do for traditional SEO still matters a great deal because it’s the foundation of discovery. But optimizing for AI-generated answers, or AEO and GEO, takes it much further than creating content for the purposes of ranking. The data created now is in service of being understood. So this means adding structured data, schema markup, and contextual attributes that are all actually teaching AI systems how products connect to real human intent.
You still need relevant keywords and inspiring product content, but now you also need semantic depth, emotional context, and data freshness so your brand shows up not just in results, but in answers.
Product content isn’t one-size-fits-all anymore. The data that makes Google perform well for you is often very different from what drives performance on Meta, or marketplaces, or retail media platforms. Each requires unique structures, attributes, and language. That means product data has to be customized by platform and constantly updated.
But it‘s important to note you shouldn’t try to “game” AI with gimmicky seasonal keywords. Instead focus on authentic, context-rich, structured data that will build durable authority. Invest in meaning not just matching.
5. Measurement will get messier before it gets clearer
Attribution as we know it is gone. Agentic shopping blurs the journey in ways traditional attribution can’t track. A shopper may discover a product via ChatGPT, verify on Google, click through directly later, and transact inside an AI interface. That path won’t generate clean numbers.
We should focus on tangible outcomes: customer visits, repeat purchases, annual spend, and incremental sales at the individual level. The signals still matter. They just won’t come with pixel-perfect certainty. This is the beginning of a new measurement era.
6. Agentic shopping is a new channel (and something bigger)
Agentic shopping will eventually sit beside paid search, retail media, and social commerce as a measurable channel, but its influence extends far beyond any line item in a budget.
It reshapes how consumers articulate needs, how AI evaluates relevance, and what it means for a product to be “discoverable.” And it rewards brands whose data is clear, trustworthy, and adaptive.
We’re not just watching a new channel emerge. We’re watching discovery itself evolve.
The New Mandate for Modern Commerce
This next era of commerce belongs to brands whose products can be understood instantly and accurately by consumers and by AI systems. That’s why AEO and data readiness deserve their own budget line, not as an afterthought to media, but as the infrastructure that makes every other dollar work harder. In an AI-led discovery world, your data is your media.
That’s why we built Lily AI: to ensure every product is enriched with the structure, intent, and clarity that modern discovery requires across every channel, every feed, every search experience, and every AI surface.
You may think agentic shopping is on the horizon, but it’s already here. And the brands preparing their product content today will be the ones customers find tomorrow.