UCP Radar: How to Optimize a Product Feed for AI Shopping Assistants

Most of the tools in this category chase AI visibility for content: blog posts, brand mentions, citations in a chatbot's answer. UCP Radar goes after a narrower and more mechanical problem, how to optimize a product feed for AI shopping assistants when that feed was written for a search engine crawler, not an agent trying to decide what to recommend.

How it optimizes a product feed for AI shopping assistants

UCP Radar connects to Google Merchant Center with an OAuth sign-in and pulls in a store's product catalog automatically. From there it runs each product against more than 50 Merchant Center compliance rules and a separate set of what it calls agentic-commerce checks, the fields an AI shopping assistant reads that a search engine mostly ignores: material, age group, standout details, FAQ-style content. Most catalogs never fill these in, because nothing forced anyone to.

The tool scores each product on a 100-point scale split across three tiers, Required, Enhanced, and Agent-Ready, then rewrites weak titles and descriptions, fills in missing fields like color, material and gender, and generates tags. Brand names are left untouched by design. The output is three feed formats: an optimized Google Merchant Center XML feed, server-rendered JSON-LD for the storefront itself, and a ChatGPT Shopping feed in JSONL. A Perplexity-facing feed is included on paid plans as well. All three regenerate automatically when the underlying catalog changes.

What UCP actually stands for

UCP is the Universal Commerce Protocol, an open standard built around the idea that product data should be structured the same way for a traditional shopping engine and for an AI agent doing the shopping. Google has been building UCP-related features into Merchant Center and its AI Shopping surfaces, and the practical effect is that a feed which merely passes Google's older compliance rules can still be invisible to an agent that's looking for structured, complete product attributes rather than a bare title and price. UCP Radar's whole pitch rests on that gap: passing Merchant Center validation and being agent-ready are not the same thing, and most stores are only doing the first one.

Pricing and the free trial

Signup requires no credit card. The trial runs 7 days, after which the account pauses rather than auto-charging, a detail worth taking at face value since it removes the usual dark-pattern risk of forgetting to cancel. Nothing on the pricing page states a product cap for the trial itself, so treat any specific number for trial-period catalog size as unconfirmed until it shows up during signup.

Paid plans start at Starter, $39 a month billed annually for 800 products, up through Growth at $99 a month for 4,000 products, Scale at $229 a month for 10,000 products with metered overage, and an Agency tier at $599 a month for 35,000 products across five Merchant Center accounts. The ChatGPT Shopping feed and server-rendered JSON-LD are gated to Growth and above, so a store on the Starter plan gets Merchant Center and Perplexity feed support but not the ChatGPT-facing output, worth knowing before picking a tier based on which AI surface actually matters for a given catalog.

Where this sits next to the editorial GEO tools already covered

AIOS Guide has already covered tools in the AI-visibility space that work on the editorial side, getting a brand cited in a chatbot's answer or tracked across AI search results. UCP Radar is not that. It never touches blog content or brand mentions. It works one layer down, on the structured product data that determines whether an individual SKU can be surfaced and recommended by a shopping agent at all. A store could have excellent editorial GEO coverage and still have a product feed that's functionally invisible to an AI assistant trying to answer "find me a waterproof men's jacket under $150," because the fields that answer that question were never filled in. The two categories of tool solve adjacent but separate problems, and a store selling physical products through Merchant Center is the specific audience this tool is built for. Anyone without a product feed, running a service business or a content site, has no use for it.

The traction, honestly

UCP Radar launched on Product Hunt this week and stood at 86 upvotes and a #17 day rank as of today. That's a real number, not nothing, but it's not evidence of a large or seasoned user base either. This is a fresh launch from a single maker, and the before-and-after examples on the product's own site (a jacket listing going from a 34 to a 92 UCP score) are demonstration cases, not independently verified customer results. The "typical uplift" stats shown on the landing page are presented without numbers attached, which is its own tell that outcome data hasn't accumulated yet.

None of that makes the underlying idea wrong. Product feeds genuinely are an under-optimized surface for AI shopping, and the gap between "passes Merchant Center validation" and "readable by a shopping agent" is real and mostly invisible to anyone who hasn't audited their own feed against it. This is a judgment pick on fit rather than a recommendation built on a track record. Worth a look if a store's product data was written years ago for Google Shopping and hasn't been touched since, worth skipping if there's no product feed to begin with.

Where it fits

For a store already running Google Merchant Center and wanting to know whether ChatGPT, Perplexity, or Google's own AI Shopping surfaces can actually parse and recommend its products, running the free 7-day check costs nothing but the OAuth connection. For everyone else, this is a category to watch rather than a tool to install today.