MarTech360 Interview with Zohar Gilad, Co-Founder & CEO, Fast Simon Inc.

“AI can multiply human capability. The unresolved question is how we develop and preserve the expertise it depends on as we delegate more responsibility to it.”

Zohar, your career has moved from engineering and product marketing at Mercury Interactive to strategy, search, and now AI-powered commerce. What early experience most shaped how you decide whether a technology is solving a real customer problem?

My first engineering job was as a research assistant in IBM’s technology division, the company’s secret weapon for tackling difficult technical challenges in large projects.

One project involved a large image bank. Memory and disk space were scarce, so we needed to reduce the amount of data required to store and display the images. Alongside compression, we reduced the image depth from eight bits to five. That saved space, but that created visible bands. The images looked like topographic maps.

A PhD mathematician on our team proposed dithering: introducing controlled noise to soften those boundaries. The eye averages across neighboring pixels, so the images looked almost as good as before, despite containing less information.

That was when I started to believe in the power of science. It also shaped how I evaluate technology: start with a real customer need, understand the constraints, and judge the solution by the experience it delivers. The customer needed usable images within the storage limits. Mathematics mattered because it made that possible.

At Mercury Interactive, you helped build categories such as Load Testing and Application Performance Management. Looking at Fast Simon today, what made you see product discovery as a category worth building rather than simply another ecommerce feature?

Great products come from combining two kinds of understanding:

1. Recognizing a technology shift and what it makes possible;

2. Knowing customers and their challenges intimately. At Fast Simon, we started with the technology.

We had a strong autocomplete technology and had seen it drive higher revenue for our media customers. That led us to ask where else it could create value.

Ecommerce was an obvious candidate. We knew very little about it at the time, but we adapted our autocomplete technology and patented our approach. The response was enthusiastic, and we found our customers: merchants and merchandisers.

That was the turning point. We moved from finding a market for our technology to understanding the problems those customers needed us to solve. Helping shoppers find products was closely connected to how merchants presented their assortment, managed priorities, and grew revenue. Search was the entry point and product discovery was the broader opportunity.

My experience at Mercury taught me that a category takes shape when you solve a meaningful, ongoing customer problem. At Fast Simon, combining our technology with an intimate understanding of merchants’ challenges helped us see an opportunity much bigger than autocomplete.

Search has evolved from matching keywords to understanding intent and context. After spending decades around search technology, what do you think the industry still misunderstands about the difference between better search and better product discovery?

The industry still tends to view product discovery as a search relevance problem. Understanding intent and context is a major advance, but it is only part of the picture.

Better search helps shoppers find what they are looking for. Better product discovery also helps them discover what they may want to buy, even when they cannot articulate or verbalize it, or cannot use the search box.

In commerce, that requires understanding both the shopper and the merchant. What is the shopper likely to want? What does the merchant want to accomplish? What is available, selling, or relevant right now? A relevant product may be out of stock in the shopper’s size, while another could meet their needs and support the merchant’s priorities. Search, merchandising, and personalization all contribute to the same experience.

The real measure is whether you help shoppers discover products they want to buy and help merchants achieve their business goals. Search accuracy contributes to that outcome, but it does not fully define it.

Also Read: MarTech360 Interview with Joep Leussink, Chief Growth Officer at AddEvent

Shopify reported that AI-referred orders grew nearly 13x year over year in Q1 2026. As AI increasingly becomes a starting point for product discovery, what does this change about how brands should think about their catalogs and merchandising?

The 13x growth statistic is significant, but we should distinguish growth from scale. It signals a change in how shoppers discover products; it does not mean AI has become the dominant channel.

I think it changes two things: First, merchants need both a human view and an AI view of the same catalog. The storefront expresses the brand through imagery, storytelling, and merchandising. The AI view needs to make products understandable: their attributes, uses, differences, availability, and suitability for particular needs.

Publishing a catalog is the starting point. I expect the next stage to involve richer dialogue between shoppers’ AI agents and merchants’ systems, such as asking questions, comparing options, and finding alternatives. That requires commerce intelligence behind the conversation.

Second, those conversations will shape shoppers’ expectations of the storefront itself. As people become accustomed to describing what they need in natural language and asking follow-up questions, brands will need to bring conversational search and AI shopping assistance into their own experiences.

Both views should reflect the same commercial reality and merchant direction: inventory, promotions, brand requirements, and business priorities. The opportunity is to make the catalog useful wherever discovery begins, while giving shoppers a compelling reason to continue the conversation with the brand.

AI can process more shopper signals than a human merchandiser ever could, but brands still need human judgment around identity, inventory, and priorities. As AI takes over more operational work, where does the merchandiser become more valuable?

The merchandiser becomes more valuable in setting direction: expressing the brand, inspiring shoppers, and deciding what the business wants to accomplish. AI is a force multiplier for that judgment.

The really good AI models understand ecommerce, such as shopper behavior, product performance, inventory, promotions, and merchant objectives. They use that understanding to optimize and personalize product discovery toward business goals, within the boundaries that the merchandisers set.

This takes much of the ongoing execution work off merchandisers’ plates. They can spend more time shaping collections, telling stories, introducing new ideas, and creating experiences that make their brand distinctive. Taste and brand identity involve human judgment, and their commercial effects are not always immediate or easy to measure.

We describe this as human expression combined with machine intelligence. Merchandisers define the experience they want to create; AI helps deliver it to individual shoppers at scale.

There is often a gap between what an algorithm predicts will convert and what a merchandiser believes the brand should promote. When those decisions conflict, how should commerce teams decide which one takes priority?

The first question is whether they are working toward the same objective. An algorithm predicting immediate conversion may conflict with a merchandiser building a new category, supporting a launch, or strengthening the brand. Those are legitimate business goals, and the model needs to understand them.

Commerce teams should set the objectives and boundaries and machine intelligence should optimize within them. Merchandisers define brand requirements, campaign priorities, inventory goals, and where editorial expression must take precedence.

Within those boundaries, I think teams should give AI room to challenge their assumptions. If a disagreement is about what will perform better, test it. If it is about a deliberate brand or strategic commitment, make that commitment explicit and evaluate it against the appropriate goal.

The merchandiser should not have to approve every ranking, and the algorithm should not define the brand. The value comes from combining clear human intent with measurable machine execution.

You’ve seen firsthand how messy retail data can be, from changing catalogs to shoppers using very different language from merchants. What separates an AI system that produces convincing answers from one that can make trustworthy commerce decisions? 

Brands need an AI system that thinks like a retailer. Generic AI models are great at languages. But ecommerce is a system. A purpose-built AI system understands inventory, new arrivals, intent, velocity, pricing, etc. And when it does, it becomes the ultimate partner to the decision makers on what and how to sell a product.

As AI agents increasingly influence product discovery; do you see them replacing the traditional digital storefront or becoming a new discovery layer around it? What do you think will ultimately determine which model wins?

I do not think AI agents are going to replace the digital storefront, especially in the fashion and accessories space. First, shopping in these categories is not simply a task to be completed. For many, it’s a fun pastime. Why would you hand over to agents something you actually enjoy?

Second, fashion is about taste, inspiration, and discovery. And visual discovery is incredibly efficient. A shopper doesn’t necessarily know how to articulate what she wants, but, thanks to the digital storefront, she doesn’t need to. Show her 20 products on a screen and, at a glance, she can spot the one that catches her eye. AI agents aren’t going to replace that.

What AI agents will do is make the storefront smarter. The future of ecommerce isn’t storefronts or agents. It’s a mixed approach. Whether the shopper is a human or an agent, the challenge will always be the same. And that’s to get the right product noticed at the right moment.

Your involvement with ClaroAI brings another perspective on how technology understands people. Its approach improves speech clarity while preserving identity and emotion. What has this taught you about where AI should intervene in human interaction and where it should step back?

For me, the lesson is that AI should help people express themselves more effectively while preserving what makes that expression their own. With speech, clarity matters, but so do identity and emotion. A technically clearer voice loses something essential if it no longer sounds or feels like the person speaking. ClaroAI’s approach reflects that distinction: help the listener understand the speaker while preserving the human qualities that make the interaction meaningful.

That gives us a useful principle: AI should intervene where it removes friction and step back where it would override human intent or individuality. The person should remain in control of what they want to communicate and how they want to be heard.

I see a similar principle in ecommerce. Merchandisers bring taste, inspiration, and brand identity. AI can handle more of the optimization and personalization, helping that expression reach the right shopper.

In both cases, the most valuable AI amplifies human expression. Its success depends as much on what it preserves as on what it improves.

You’ve watched technology move from enterprise software and search to personalization and AI agents. As AI shifts from helping people make decisions to acting on their behalf, what part of today’s AI narrative do you think we are getting wrong?

I think we are confusing the ability to act with the judgment to know what should be done. An AI agent can execute, but it still needs clear objectives, context, and an understanding of tradeoffs. In commerce, “sell more” means different things depending on whether the merchant is building a brand, protecting margins, or clearing inventory.

But there is a deeper challenge: we assume humans will provide the judgment while AI does the work. Human judgment comes largely from doing that work, including making mistakes and learning from consequences. If AI removes the lower rungs of the career ladder, how will the next generation develop expertise? And how will experienced professionals retain the skills needed to evaluate AI’s decisions?

AI can multiply human capability. The unresolved question is how we develop and preserve the expertise it depends on as we delegate more responsibility to it.

Thanks, Zohar!