Rushify AI
Run free check →
How Rushify works

11 signals.
12 platforms.

No black boxes. No proprietary scoring. Here is exactly what we check on your site, who we test against, and how we turn it into a number out of 100 and a grade between A and E.

01 · What we check

The 11 signals AI agents use to read a business.

Each signal is checked against your real homepage in real time. Weights below are out of 100 and sum to 100 exactly.

#
Signal
Why it matters
Weight
01
HTTPS https://
First trust signal. Sites still on HTTP get skipped by most modern AI agents entirely.
8
02
robots.txt /robots.txt
First file an agent fetches. Clear crawl rules tell AI agents you're a real, indexable business.
8
03
AI policy ai.txt or llms.txt
Emerging standards. ai.txt and llms.txt let you state explicitly that ChatGPT, Claude, Perplexity and others are welcome. Either passes.
8
04
Sitemap /sitemap.xml
A valid sitemap means agents discover every page, not just your homepage. Without it, deep content stays hidden.
6
05
Structured data JSON-LD
Schema.org markup is how agents understand what your page is about. The single biggest fix for most SMEs.
16
06
Page title and description
Title tag and meta description. Tell AI agents what each page is about, and they appear in search results too.
10
07
Open Graph tags
og:title, og:description, og:image. Cheap to add. Used by agents for previews and disambiguation.
6
08
Canonical URL rel="canonical"
Tells agents which URL is the real one. Without it, multiple page variants get treated as duplicates and ignored.
8
09
Heading structure H1, H2
One clear H1, supporting H2s. Agents use these to understand what each part of your page is about.
10
10
Language declared <html lang>
Says what language the page is in. Agents may otherwise serve it to the wrong audience or skip it.
6
11
Image alt text
AI agents read alt text to understand what's in your photos. Critical for product images especially. Highest single weight after structured data.
14

Weights sum to 100 exactly. We re-tune the rubric quarterly based on observed citation patterns. The rubric is public on purpose: no black box, no vibes-based grading.

02 · Who we test against

Five global AIs.
Plus the APAC models that matter.

Five global assistants get tracked for every Pro+ customer. The APAC platforms are the consumer models your buyers actually ask in their own markets - you track whichever ones matter to you.

Global · The five assistants tracked every week for Pro+

GP
Western · Global
ChatGPT
OpenAI

800M weekly users. Largest single AI buyer-discovery surface. ChatGPT Go launched across Asian markets through 2025.

CL
Western · Global
Claude
Anthropic

Used for business research and product comparison. Strong fit for B2B and considered-purchase SMEs.

PX
Western · Global
Perplexity
Perplexity AI

Tens of millions of queries per day. Closest replacement for Google search behaviour. Buyers ask, get a single answer.

GE
Western · Global
Gemini
Google

Bundled into Google Search via AI Overviews. Effectively unavoidable, and where SMEs lose clicks first.

MA
Global · Asia-heavy
Meta AI
Meta

Built into WhatsApp, Instagram and Messenger - the apps Southeast Asia lives in. India is its single biggest market. For many buyers, the assistant they use without thinking of it as one.

APAC · The consumer models your buyers ask

DB
Doubao 豆包

ByteDance's consumer assistant and China's most-used. Integrated with Douyin and Toutiao.

YB
Tencent Yuanbao 腾讯元宝

Built into WeChat, where over a billion people already are. The default for the Tencent ecosystem.

QW
Alibaba Qwen 通义千问

Alibaba's flagship assistant. Strong on writing and general queries, consumer and business.

DS
DeepSeek 深度求索

China, Hong Kong, and broadly used across APAC. A strong entry point for the China market.

B
Baidu ERNIE 文心一言

Particularly strong on search-style queries in mainland China.

SL
SEA-LION AI Singapore

Multi-country SEA: Burmese, Filipino, Indonesian, Khmer, Lao, Malay, Tamil, Thai, Vietnamese.

SA
Sahabat-AI Indosat + GoTo

Bahasa Indonesia, plus Javanese, Sundanese, Balinese, Batak. Indonesia's national AI.

The price stays USD 12.50 per URL whatever you track. We add platforms as they prove out as genuine consumer surfaces, and a social discovery layer - how you surface across the social and commerce platforms your customers use, and the AI recommendation engines running underneath them - is in development for Pro+. Pro+ customers vote on what we add next.

03 · How we score

From 0 to 100.
A to E. No tricks.

Each signal contributes points based on its weight and whether it passes, partially passes, or fails. We sum the result and round to a whole number.

A
85 to 100

Agent-ready

You appear consistently in AI answers for your category. Keep monitoring; the bar moves.

B
70 to 84

Good · keep going

Most signals firing. One or two structural fixes will push you into the top three for your category.

C
50 to 69

Mixed signals

Agents can find you but don't always recommend you. The fixes report tells you exactly what's missing.

D
30 to 49

Needs work

You're losing queries to better-prepared competitors today. Most fixes are afternoon-sized.

E
0 to 29

Mostly invisible

Your site is functionally invisible to AI agents. The good news: there's nowhere to go but up, and it's not expensive.

What we do

  • Fetch your homepage live and parse the actual HTML, robots.txt, sitemap.xml, ai.txt, llms.txt
  • Score against a published rubric, re-tuned quarterly
  • Show our working: every check, every grade, every fix
  • Sort fixes by impact, not by what's easiest for us
  • Treat APAC platforms with equal weight to Western ones

What we don't

  • Store your URL or email after the check is rendered
  • Sell, share or resell data to third parties
  • Use opaque ML models for scoring (it's a rubric, not a vibe)
  • Penalise you for not being on Western platforms
  • Hide the rubric, the proxy code, or the prompts behind a paywall
Your turn

Run your check now
in 60 seconds.

https://