Beyond Specs: How the Ecom Data War Moved to Conversational AI

There was a time when selling products online was simple. You had a product, you had a spec sheet, and if you were a retailer, you paid a company like CNET Content Solutions or Etilize a handsome subscription fee to fill your webshop catalog. Then the world changed.

First, Icecat flipped the script with Open Icecat, charging brands instead of merchants and giving standard specs away for free. The legacy players got swallowed up by enterprise conglomerates, PE-funded PXM platforms moved in with six-figure SaaS price tags, and everyone settled into a new normal.

Fast forward to today. The “new normal” just got smashed by genAI, compressed SaaS valuations, and a whole new way people shop online. Here are the four structural shifts reshaping product data and why the quiet battle over catalog feeds just became a high-stakes war for consumer engagement.

1. The Death of Gated Catalog Feeds

Remember paying to access standard product databases? Retailers certainly do, and they hated it. When Open Icecat gave merchants high-quality, manufacturer-verified datasheets for free, the two-sided network effect took over. Brands paid to ensure their gear was mapped accurately everywhere, and merchants happily ditched paid feeds to save their margins.

The legacy subscription model collapsed. Incumbents were quietly absorbed into giant enterprise analytics engines – CNET into 1WorldSync, Etilize into GfK and NielsenIQ – where they were repurposed for backend supply chains rather than open-web retail.

The takeaway? Information wants to be free, but brand syndication wants to be funded. As a consequence Open Icecat become globally the dominant product content database and syndication network. Every day pushing gated catalog feeds further to the periphery of the ecom market.

2. The Great SaaS Squeeze & The PXM Panic

With simple catalog feeds solved, enterprise Product Experience Management (PXM) vendors like Salsify and Syndigo showed up. They built massive SaaS businesses selling workflow governance, localized text variants, and retailer mapping rules. Life was good. Although Icecat started to nibble away the margins with its unique AI-powered taxonomy competences and low-cost or even free PIM/DAM mapping solutions. Adding supply chain integration (EDI) into the mix for transaction oriented merchants.

Then came high interest rates, tight software budgets, and Large Language Models. Suddenly, a mid-level LLM could reformat, translate, and map unstructured product copy across thousands of channels for a fraction of a cent. Paying $150k a year for glorified text translation and field mapping suddenly felt like buying a luxury sports car just to commute two blocks.

To survive the “SaaSpocalypse,” the PXMs are desperately pivoting upstream or selling out to private equity for a fraction of their hype values. At the same time, they are embedding autonomous AI agents to handle real-time conversion tracking and closed-loop product detail page optimization. They aren’t just storing your data anymore, they are fighting to prove they can actively drive your conversion rates. But does that work if they are actually not part of the conversation?

3. Downstream Power Moves: Entering the Studio

While PXMs try to defend their enterprise software turf upstream, the downstream experience is where the consumer actually pulls out their credit card. Plain bullet points don’t sell premium gear: visual experiences do. That’s why the expansion into Icecat Studio for Product Stories is so strategic. By stepping into Flixmedia’s domain, Icecat moves directly onto the store page.

Why does this matter? Because while AI can write ten thousand product descriptions in five seconds, it can’t magically create brand-approved 3D models, hot-spot asset layouts, or interactive video widgets. By embedding dynamic, single-script visual components right below the fold, you bypass slow retailer IT queues, protect rich media IP, and drive measurable revenue lift right at the point of decision.

4. Meet LENA: Grounded AI Meets Shopping Intent

The final shift is where things get truly interesting. Traditional e-commerce search is broken. We’ve all been trapped in parametric filter hell, clicking endless check boxes for “RAM,” “Voltage,” or “Dimensions,” only to get zero results. As buyers switch to conversational AI assistants, search changes completely. But generic LLMs have a fatal flaw: they hallucinate. If an AI shopping bot tells a B2B buyer that a server power supply fits a rack when it doesn’t, someone gets fired.

Enter LENA.

Because LENA is trained directly on structured, manufacturer-verified catalog data, it offers deterministic, zero-hallucination answers. It uses Icecat’s MCP server to retrieve grounded Icecat product data.

Traditional Web Search	Conversational Discovery with LENA
Typing keywords into search bars	Asking natural questions (“Will this fit my 2U rack at 240V?”)
Scrolling through 40 flat PDP tabs	Getting a grounded, instant recommendation
Optimizing for traditional Google SEO	Optimizing for Generative Engine Optimization (GEO)

If a buyer can simply ask LENA what product solves their exact problem, the search box dies. The platform that provides the verified answers becomes the new front-end of e-commerce. LENA is still in her infancy. Many features need to be further developed and more MCP server tools will be added before it has the required versatility to outcompete its older hallucinant siblings.

Towards Conversational AI

Product data isn’t just about back-end hygiene anymore. The game has evolved from storing specifications to driving conversion. By pairing an open spec pipeline with dynamic visual storytelling (Studio) and zero-hallucination AI intelligence (LENA), the humble catalog database is becoming the central engine of conversational commerce.

Classic ecom search is dead. Long live the conversational AI experience.