Validate your website for AI agent commerce using UCP (Universal Commerce Protocol) compliance testing.
Last updated
Agentic Optimization tests whether your e-commerce store is ready for AI agent commerce. AI agents like ChatGPT, Gemini, and Copilot are increasingly able to browse, compare, and purchase products on behalf of users. The Universal Commerce Protocol (UCP) is the standard that enables this.
UCP is a protocol that allows AI agents to discover, interact with, and complete purchases on merchant websites. It works through a machine-readable profile at /.well-known/ucp.json that describes your store's capabilities, API endpoints, and checkout flow.
Without UCP compliance, AI agents cannot browse your product catalog or process orders on your site.
Agentic Optimization is a Pro+ beta feature — it requires a Pro, Ultra, or Enterprise plan and beta access (request it from the feature page). UCP validation is configured at 3 credits per run, but the action is currently in shadow mode — no credits are deducted on any tier, and the page does not display a cost before you run. Past validations are kept in the history list below the form.
The feature has four tabs:
| Tab | What it does |
|---|---|
| UCP Readiness | Validate UCP compliance and review validation history |
| Product Catalog | Import and manage the products you monitor in AI answers |
| Shopping Journey | Product visibility across the awareness, consideration and decision stages of an AI shopping journey |
| Brand Knowledge | Shopper personas, competitors, and markets that shape prompt generation |
| Tab | What it does |
|---|---|
| UCP Readiness | Validate UCP compliance and review validation history |
| Product Catalog | Import and manage the products you monitor in AI answers |
| Brand Knowledge | Shopper personas, competitors, and markets that shape prompt generation |
yourstore.com) — any URL on your store's domain works; the validator locates your UCP profile at /.well-known/ucp automatically| Type | Duration | What it checks |
|---|---|---|
| Quick Scan | ~30 seconds | Fetches and validates your ucp.json profile structure, required fields, and capability declarations |
| Full Audit | ~60 seconds | Complete AI agent simulation including discovery, API access, checkout flow, and payment readiness |
Start with a Quick Scan to confirm your UCP profile exists and is correctly formatted. Use Full Audit for a comprehensive assessment that simulates the entire agent interaction.
The Full Audit produces an overall score (0–100) broken into four categories.
Can AI agents find your UCP profile? Tests profile accessibility at /.well-known/ucp.json and metadata completeness.
Can AI agents interact with your store's API? Tests REST endpoint responsiveness, schema validation, and sample operations.
Can AI agents complete a purchase? Tests checkout session creation, order flow support, and fulfillment capability.
Are payment handlers correctly configured? Tests signing key validity, webhook verification, and payment handler availability.
Each category shows a progress bar with its score. Bars are color-coded:
| Color | Score range | Meaning |
|---|---|---|
| Green | 70%+ | Passing |
| Yellow | 40–69% | Partial compliance |
| Red | Below 40% | Failing |
Your overall score maps to a letter grade.
| Grade | Score | Meaning |
|---|---|---|
| A | 90–100 | Fully AI-agent ready |
| B | 80–89 | Good — minor improvements needed |
| C | 70–79 | Fair — moderate issues to address |
| D | 60–69 | Poor — significant problems |
| F | Below 60 | Failing — critical issues blocking agent commerce |
The results show each step the validator executed with its status:
Each step shows the time it took in milliseconds, helping you identify performance bottlenecks.
Below the results, actionable recommendations tell you exactly what to fix. Examples:
| Error code | What it means |
|---|---|
UCP_MISSING_ROOT | No UCP profile found at /.well-known/ucp.json |
UCP_ENDPOINT_NOT_HTTPS | Endpoints must use HTTPS |
UCP_MISSING_SERVICES | Required service definitions are missing |
UCP_MISSING_SIGNING_KEYS | Order capability requires signing keys |
UCP_SCHEMA_FETCH_FAILED | Could not fetch the remote schema |
The Product Catalog tab tracks the specific products you want to monitor inside AI answers.
Click Import from your site and enter your store URL. Vizzybl detects your platform (Shopify, WooCommerce, BigCommerce, Magento, or custom), reads public catalog data and schema.org Product markup, and discovers your products — no store connection or API key required. The import:
Discovered products and categories land as suggestions — confirm, edit, or dismiss each one. Re-running an import updates prices and quality scores but never overwrites your edits or resurrects items you dismissed.
Product limits by plan: Pro 25, Ultra 100, Enterprise 300. Imports that find more than your limit keep the most prominent products and tell you how many were left out.
Use Link prompts to connect monitored prompts to the products they reference — Vizzybl suggests links by matching product names and aliases against your prompt texts, and you confirm which to apply. Linked prompts power product-level visibility metrics (rolling out in a future phase).
At the bottom of the Shopping Journey tab, the Google AI Mode shopping picks table lists every product Google put inside its AI Mode answers to your prompts in the selected window (7, 30 or 90 days). It answers one question: when a shopper asks Google AI Mode what to buy, are your products in the picks, and why does Google prefer the ones it chose?
| Column | What it shows |
|---|---|
| Product | The product as Google named it, and the prompts it was picked on |
| Seller | The shop Google listed it from |
| Times picked | How many of your AI Mode answers showed it |
| Avg. position | Where it sat among the products in the answer (1 = first) |
| Latest price | The price Google showed most recently |
| Google's reason | Google's own words on why it picked the product |
| What to do | Your next step, with a button where there is one |
What to do depends on whose product it is:
A line above the table gives the denominator: how many of your AI Mode answers in the window showed products at all. Shopping picks are collected from the day they are switched on for your workspace; earlier answers have none. The table is part of Ultra and Enterprise, the plans that monitor Google AI Mode; other plans see a preview.
Shopper personas describe who buys from you and how they shop — a price-conscious student and a premium-quality enthusiast phrase their questions to AI assistants differently. Click Suggest personas to seed personas from your brand analysis, then edit names, descriptions, regions, and interests. Personas (together with your markets and competitors) condition catalog-driven prompt generation.