> aihacked.com :: answer-box autopsy
We open the answer box
and show you the trick
AI Overviews, ChatGPT, Perplexity, Claude, and Gemini decide what buyers see before they ever reach your site. AI Hacked tears apart the pages that win those answers — and hands you the schema, the sentence shapes, and the counter-spells.
- Teardowns live
- 3
- Tactics catalogued
- 6
- Surfaces graded
- 8
The library
Latest teardowns
How Stripe owns the AI Overview for 'best payment processor for startups'
It's not the brand. It's the docs page Stripe pretends is marketing.
Why ChatGPT cites Reddit instead of your product page
The model trusts the consensus, not the brochure. Here's the loophole.
The press release engineered to be cited by Perplexity (not read by humans)
70% of news queries get LLM-answered. So Sequoia stopped writing for journalists.
The thesis
The SERP isn't the SERP anymore.
An estimated 30–50% of Google searches now resolve in an AI Overview. ChatGPT, Perplexity, and Claude answer billions more without ever showing a blue link. The page that gets cited wins. The page that doesn't might as well not exist.
AI Hacked is the running record of which pages win — and why. Same teardown DNA as Marketing Hacked and Sales Hacked. New arena, same magic.
Audit your brand across ChatGPT, Gemini, and Perplexity. See every cited URL, snippet, and competitor — with a weekly auto-rerun.
AI Summary Optimization for Apple Intelligence and Gmail Gemini. Win the 1-sentence preview before the open.
The wire is dead. Grade your release for the agent that now reads it — JSON-LD, atomic facts, citeable quotes.
Be the answer when a recruiter prompts an LLM. Resumes turned into structured, citeable profiles.
The marketing industry's running record of who called what. Predictions, scored against what actually happened.
The greatest ads, landing pages, and emails — hacked open, trick by trick. The original teardown library.
The encyclopedia
Tactics catalogue
Every trick we see in the wild, indexed by category and rated by magic color. Start here if you want the playbook before the autopsies.
Content
Atomic Facts
Write one fact per sentence. LLMs extract sentences, not paragraphs.
Schema
JSON-LD Coverage
If it's not in schema, the model has to guess.
Distribution
llms.txt
robots.txt for the LLM age. A curated map of your most citeable content.
Content
Citeable Quotes
Named, attributable quotes are the highest-weighted text type in LLM citations.
Content
Named Entity Density
Name people, products, places, and competitors. Vague pages don't rank as anything.
Authority
Consensus Engineering
LLMs cite Reddit, YouTube, and HN for consideration-stage queries. Show up there honestly.