Zyntent Product Update — September 2026

In this update: Widgets/Brand Search/API/Coming later this month

As more product discovery moves into chat, AI search, and on-page assistants, the moment that used to end at a search results page now happens inside the content itself. Zyntent’s job is to catch that moment in the form of a query, a prompt, or an intent-carrying phrase, and match it to something real, in real time, without turning the page into an ad unit. This update covers three things we shipped this cycle, plus what’s next.

Widgets

We shipped three widgets, and the important part isn’t that they exist – it’s how they’re different from what came before.

Widget v1 summarized a page and appended a product block at the end. 

Widget v2 does something structurally different: it scans the full article text, identifies the words that carry commercial intent — a product name, a category, a comparison — and inserts links inline, annotating the existing copy instead of adding a new element. Same page, no new UI block. The text itself becomes the placement.

Alongside it, a storefront widget: a publisher writes a natural-language prompt, and Zyntent returns a curated set of roughly 100 relevant products, without manual catalog curation and category-feed configuration. Useful for content creators and affiliate publishers building product pages from scratch.

Brand Search

Here’s the story, plainly: a single-word brand query — just “Nike,” no product type — used to return zero results. Publishers send this kind of query at real volume: no category, no model, just the brand name, and all of it went unserved.

The fix: brand-only queries now return every matching product regardless of product type. That closes a gap worth roughly 10% of total query volume, and the share of queries that contain only a brand name and nothing else.

Separately, we shipped per-brand alias resolution across all 12 supported languages — a per-brand list of known alternate forms, covering abbreviations and partial names (“DG” and “Dolce” both resolve to “Dolce & Gabbana”) and common punctuation/formatting variants (“H&M” typed as “HM” or “H and M”).

Worth being precise about what this is and isn’t: it’s a coverage fix, not an accuracy fix. These queries went from 0% served to 100% served. Match accuracy within them is a separate metric we don’t have a clean benchmark for yet.

API

Zyntent now offers a search API that returns product matches with no Zyntent tracking link attached. A technology partner applies their own deep link and keeps their own attribution.

A concrete case: a publisher running their own Amazon Associates account uses Zyntent’s intent search to find matching Amazon products, then swaps in their own deep link. Zyntent is the search layer; the publisher monetizes on their own account, however they choose to.

The business model here is different from the rest of the platform, on purpose: we’re not taking a share of the resulting sale, we sell API call tokens. It’s not in the public API yet, and it’s not a self-serve flow. If it’s useful to your stack, contact Zyntent team and we’ll set it up.

Coming later this month

Two things already in motion, not shipped yet: product search by specific model name (“iPhone 15 Pro Max,” not just “phone”), and a jump in active language support from 6 to 12. Aliases are already built for all twelve, this is about activating the remaining six, alongside product catalog coverage for new markets including Turkey, India, and Lithuania.

See it working: find widgets and search live in the sandbox at Zyntent. You can access it without integration on live data. For the API, or a walkthrough on your own content, get in touch or start with the docs.

Contact me on Linkedin to talk about monetizing intent on your platform.

Kyrylo Kravchenko

Product Lead