AI Visibility — agent flow

Test harness · Stage A → B → C → D.

Set up a brand to track

Fill the required fields; leave targeted customer and price blank to let the agent research them from your products + category. It also infers rival SKUs, languages, and prompt count — you review everything next.

Brand

Brand name

Website URL (the most reliable brand identifier)

Product

Category (required)

Key products (required, comma-separated)

Customer & pricing

Targeted customer (optional — researched from products + category if blank)

Price positioning (optional — researched from products + category if blank)

Market

Markets (required, comma-separated)

Known competitors (required, comma-separated)

AI models

Fan out across (tick the models to query)

Western modelsanswer every language

ChatGPT (GPT-5.6 Luna)OpenAIGemini (3.6 Flash)Google

Chinese modelszh-Hans / zh-Hant only

Qwen (3.8 Max)AlibabaDeepSeek (V4 Pro)DeepSeek

Prompt mix

Sample split (how the prompt count is divided across the funnel — weights, normalised to 100%)

Branded15%

Names your brand, so you see how well people who already know you recall it.

Category35%

Names the product type but not your brand, so you see whether you show up among the “best X” picks.

Problem50%

Names only the need, so you see whether your brand comes up when nobody asked for it.

Default 15 / 35 / 50 — weighted toward unbranded discovery.