Platform AI Radar
Ask real domestic and global AI engines on a schedule; monitor mention rate, ranking and source composition of your brand, products and key entities — accumulated into a trackable visibility profile.
Separate profiles per engine: mention rate, ranking, recommendation rate and top-answer rate over time, with trends you can revisit.
Definition: every rate carries its denominator; engines without samples show "no data yet", never a misleading 0%.
Your brand and competitors side by side on the same questions — see exactly which questions and which engines create the gap.
Which websites AI answers cite, how much of it is your own content, and which layer sources come from (owned / media / community).
Identify visits from 16 named AI crawlers to your website, and compare against robots.txt — flagging crawlers you meant to block that got through, and those you meant to allow that never showed up.
Sudden mention drops, negative descriptions or declining description accuracy trigger alerts, with archived evidence.
The same question set is retested periodically to compare before/after changes; when sampling conditions (engines / models / settings) change, results are marked "not comparable" instead of forcing a curve.
9 China engines covered:
6 global engines:
Names and counts are computed from the engine registry — the same source as the homepage stats band, the free check and plan engine ranges.
These four panels are the core views of the tenant "AI visibility" workspace (demo data here). Real numbers come from sampling on your own projects: every chart carries denominators and sample sizes, with definitions identical to the console.
| Engine | Samples | Visibility | Top 1 | Top 3 |
|---|---|---|---|---|
| 豆包 | 128 | 41.7% | 12.5% | 38.3% |
| DeepSeek | 116 | 38.2% | 10.3% | 35.8% |
| Kimi | 104 | 35.5% | 9.6% | 33.1% |
| 腾讯混元 | 92 | 29.8% | 7.7% | 27.6% |
| 通义千问 | 84 | 27.4% | 6.9% | 24.9% |
| 百度 AI 搜索 | 76 | 24.6% | 5.8% | 22.3% |
| 智谱 GLM | 64 | 22.1% | 5.2% | 19.8% |
| 文心一言 | 58 | 18.6% | 4.1% | 16.2% |
| 腾讯元宝 | 52 | 16.3% | 3.4% | 14.7% |
| OpenAI | 44 | 13.8% | 2.6% | 11.9% |
| Perplexity | 38 | 11.2% | 1.9% | 9.6% |
| Gemini | Engine not covered in this sampling group | |||
| Claude | Engine not covered in this sampling group | |||
| Grok (xAI) | Engine not covered in this sampling group | |||
| Mistral | Engine not covered in this sampling group | |||
Engine names match the live registry; the values are demo. Real definition: visibility is per-engine over its valid samples, and an engine that was never sampled shows "no data" rather than 0% — "not measured" and "measured at zero" are opposite findings.
The center line is weekly visibility; the band is the rolling 7-day range (demo). Single snapshots mislead: the same question gets different answers on different engines at different moments — so you read trends.
Three layers are three independent layers, not a funnel: mentioned by AI ≠ cited by AI ≠ visited from search. Numerators and denominators are disclosed per layer in the console.
Bars show brand visibility; the gray band is an industry reference range (P25–P75, demo). When the baseline sample is insufficient it is downgraded to "for reference only" — no conclusions.
Definitions match the console exactly: denominator = valid samples, zero-sample shows "no data", mixed capture methods never merge into one trend line. All figures here are demo; sign in to see your project's real data.