I’ve been doing SEO since before Google existed, and I’ve watched multiple waves of disruption reshape how brands are discovered. But my conversation with Aniket Deosthali, of Envive AI, highlighted a shift that feels bigger than most: we’re no longer optimizing only for human shoppers. Increasingly, we’re optimizing for agents.
Envive makes storefronts “self‑improving” – learning from every interaction so they sell a little better each day. The twist is that the “shopper” is often no longer a person scrolling through pages, but an AI agent like ChatGPT or Gemini acting as a broker or personal shopper. The brands that win will be those that serve both human visitors and AI agents at the same time.
Beyond “People Who Bought This Also Bought That”
For years, “wisdom of the crowds” powered recommendation engines: if enough people who bought product A also bought product B, then B was pushed as a cross‑sell. That works, but it demands lots of data and fails when you introduce new products or want more nuance around intent.
Envive’s answer is an “AI brain” for the storefront. In Aniket’s cookware example, if a shopper adds a non‑toxic pan to their cart, the system doesn’t just surface random accessories. It reasons about why the shopper chose that product. Someone willing to pay a premium for non‑toxic cookware probably also cares about non‑toxic storage containers for leftovers. That becomes a chance to present a contextual recommendation with a narrative: you protected your family at the stove; here’s how to extend that to how you store food.
That shift from simple correlation to reasoning is profound. It turns recommendations from generic product shoves into meaningful suggestions with “reasons to believe.” And that, in turn, is how you grow average order value and revenue per visitor in a way that feels like help, not pressure.
Designing for Humans and Agents
Most e‑commerce sites today are still optimized almost entirely for human eyeballs: hero images, branding, lifestyle photography, video. Agents don’t see any of that. They see the structured data, product attributes, copy, content architecture, internal linking, and the user‑generated content around your brand.
That creates a dual‑optimization challenge. Just as the mobile revolution forced everyone to rethink sites for smaller screens, the AI shift forces us to build a parallel layer for agents. You still need to delight humans on your site, but you also need to ensure agents understand what you sell, who you serve, and why you’re the best answer.
In many ways, this is SEO 2.0 – or GEO, generative engine optimization. Instead of a few keywords in a search box, you now face long, conversational queries with rich context and refinements. The long tail becomes far more important. Your first‑party data and content must be expressed in a way that’s “robot‑friendly” and ready for vectorization, so agents can map your brand to the right needs.
Meritocracy, Feedback Loops, and Truth
A central theme in my discussion with Aniket is AI‑driven meritocracy. In the past, you could sometimes out‑market a mediocre product for a long time. Now, every touchpoint becomes a data point. Positive experiences generate reviews, posts, and UGC that feed the models. Negative experiences do too.
No amount of ad spend can sustainably hide a poor product when systems constantly ingest authentic feedback and use it to rank, recommend, and respond. That cuts both ways: brands that genuinely delight a well‑defined audience will see those signals amplified, especially if they actively capture first‑party data and user‑generated content and place it where agents can find it.
Preparing for the Next Year
If I distill Aniket’s advice for DTC marketers and e‑commerce leaders into one idea, it’s this: deeply understand your personas and use cases, then make those truths legible to large language models.
Many brands already segment audiences for media and creative, but their sites still present a single, generic story. The opportunity is to align product, content, and data around real personas, then structure it so agents can see not just what you sell, but who you sell it to, why it works, and how it performs in the real world.
LLMs and agents will reward truth and performance. We are still early in this shift, which means first movers who get their data, content, and storefronts ready for both humans and agents can build durable advantages that slower competitors will struggle to overcome.
Discussion points include:
- How storefronts can become self‑improving systems that serve both human shoppers and AI agents
- Why moving beyond “people who bought this also bought that” toward reasoning‑based recommendations is critical for driving AOV
- What it means to design e‑commerce experiences simultaneously for human eyeballs and machine readers
- How GEO and long‑tail, robot‑friendly content are reshaping discoverability
- Why AI‑driven feedback loops are turning e‑commerce into a meritocracy where truth, performance, and product quality ultimately win