How to Tell If You’re Buying a Wrapper or Real AEO Infrastructure
I dug into ZeroRank, Erlin, Visby, and Blazly - and the architecture gap between them is not subtle.
ZeroRank is the only one running what looks like real scraping infrastructure. The other three are workflow wrappers with nice dashboards, built on public developer APIs.

Here’s how to spot the difference:
The checklist:
1. Engine coverage. If a tool only tracks ChatGPT, Gemini, Claude, and Perplexity - wrapper. All four have clean public APIs. The real tell is Google AI Overviews and Bing Copilot. No public API exists for either. If a tool skips both, it’s because they can’t scrape them.
2. BYOK. If a tool lets you plug in your own OpenRouter or LLM API key for tracking, that’s a confession, not a feature. They’re routing prompts through your key to the same public endpoints you could hit yourself for free. Zero proprietary data layer.
3. Pricing math. Real scraping infrastructure - proxies, rotating IPs, headless browsers, anti-bot evasion - costs real money. If a tool offers unlimited tracking at $50-100/month, the math only works if their marginal cost per query is near zero. Which it is, if they’re just calling APIs.
4. Rendered screenshots. Real scrapers can show you an actual screenshot of a Google AIO result. Wrappers only have text. Ask support. A real infrastructure player answers immediately. A wrapper has nothing to show you.

The verdicts:
ZeroRank - Covers AIO and Bing Copilot. No API route exists for those. You either scrape or you don’t cover them. Real infrastructure, or at minimum the only tool with evidence of it.
Erlin - Four engines, all public APIs. Accepts your own OpenRouter key for tracking. That’s not a data product, that’s a prompt runner.
Visby - Three engines, all API-accessible. Founder confirmed in an AppSumo Q&A: “We use official LLM APIs to execute prompts - we do not track through the public web apps.” Wrapper confirmed by the founder himself.
Blazly - Content and GEO optimization tool with tracking bolted on. Same API-only engine coverage. Workflow tooling, not a data engine.

One honest caveat - wrapper doesn’t automatically mean worthless. For pure LLM engines like ChatGPT and Gemini, API responses and consumer UI responses are largely the same. The real blind spot is AIO and Copilot specifically. That’s where wrappers fall short and where it actually matters.