See Who Wins AI Search in Your Niche (Claude Code Agent)

A Google result gave everyone one ranking. An AI answer gives your buyer a ranked shortlist the engine already decided to trust, and every engine trusts a different set. Your category has a different number one on ChatGPT than on Perplexity, and most companies have never checked a single board, let alone all five.

This is the exact leaderboard agent we built live in Claude Code. It runs your buyers' real questions across ChatGPT, Perplexity, Gemini, and Claude, counts who gets named, and builds a ranked leaderboard per engine with every brand's share of voice. It measures the real answer. It never asks the engine to rank your category from memory, because it will invent a plausible order if you let it.

It's the agent we built live, packaged so you can install it and rank your category in about 25 minutes.

What's in the download:

• The full Claude Code project, with a README and build steps that include every prompt

• The agent's operating rules (the measure-don't-rank discipline)

• Three reusable prompts (run and record, rank and tally, recovery) and a leaderboard template

• A prompt-set template and a filled example leaderboard so you can see the shape

Built by Will Leatherman and the team at Catalyst, who've built AI-search systems for 150+ B2B companies. Every rank traces to a saved answer, so your share of voice is a real number, not a guess.