AEO/GEO intelligence for decentralized ecosystems. Know exactly what every AI says about your protocol.

Coming soon
Joined August 2026
We measured where ChatGPT, Claude, Gemini, Grok & Perplexity get their information about @Rocket_Pool. The #1 cited domain is reddit.com — cited in 16.8% of answers. That's ahead of rocketpool.net itself (12.2%). 🧵
1
1
88
Reddit is cited in 16.8% of AI answers about ETH liquid staking. The protocol's own website: 12.2%. If you're not writing the Reddit thread, someone else is writing your AI answer.
1
25
Where @Rocket_Pool disappears from AI answers isn't random. Trust & safety questions: 20% omit it. How-to questions: 6.7%. Models hedge exactly where users worry the most.
1
6
Same 50 prompts about ETH liquid staking. Gemini mentions @Rocket_Pool on 50/50. Claude: 38/50. A 24-point gap on identical questions. Which AI your users ask determines what they learn about you.
1
41
The gap is concentrated in one place: trust & safety prompts. 20% of answers to risk and safety questions omit Rocket Pool, vs 6.7% on how-to prompts. Worst case: "What are the biggest risks of liquid staking derivatives?" 1 engine out of 5 mentions it.
1
14
We're building @tryalveon to track this weekly. Follow for the next report.
5
First AI Visibility Report drops Tuesday. 50 prompts × 5 AI engines on one of Ethereum's biggest liquid staking protocols. 500 archived answers, 2,136 citations. The #1 cited source isn't the protocol's website. ⏳
1
24
Building Alveon in public, note #1: We measure what 5 AI assistants say about a crypto protocol, 50 prompts, 2 signals each, every answer archived. First public report drops Sept 1. It's about the second-largest ETH liquid staking protocol, and the results surprised us.
1
45
We've been running thousands of prompts through ChatGPT, Claude, Gemini, Grok and Perplexity to see what they actually tell users about crypto protocols. They're confidently wrong more often than you'd think. Weekly reports start soon.
4
4
98