Context
The setup before I touched it.
Buying behavior is quietly migrating into LLMs. Instead of Googling 'best CRM for small teams' and clicking three blue links, people now ask ChatGPT, Perplexity, or Gemini and get a single recommendation. That recommendation rarely matches the SERP - and brands have no idea whether they're being recommended, ignored, or actively trashed. SEO tools watch the wrong layer.
How it came together
Step by step - expand any phase for the highlights.
- Mapped LLM-driven discovery behavior
- Compared SERP results vs LLM answers
- Quantified the visibility gap for brands
Problem → Approach → Result
The short version, for the broad audience.
Problem
- 01There is no clear way to know whether AI assistants are recommending your product for the prompts that actually drive purchase intent.
- 02Citation sources differ wildly between models (ChatGPT cites Reddit, Perplexity cites docs, Gemini cites whatever) and shift week to week with no warning.
- 03Marketing teams pour budget into SEO while a measurable share of buying conversations has already moved into LLMs they don't measure.
- 04When a competitor suddenly dominates an LLM's recommendations, you currently find out from a sales call, not a dashboard.
Approach
- 01Define a tracked set of buyer-intent prompts per brand and category - the exact phrasing real customers use.
- 02Run scheduled probes across multiple AI models (ChatGPT, Perplexity, Gemini, more) and capture mentions, position in the answer, and every cited URL.
- 03Score share-of-voice and surface week-over-week movement at a glance - so a competitor surge is a notification, not a surprise.
- 04Show which sources LLMs are pulling from so content teams know exactly which Reddit thread, comparison post, or docs page to influence.
Result
- 01Brands get a measurable AI-visibility metric instead of guessing whether they're being recommended.
- 02Content and PR teams know exactly where to invest - influence the cited source, move the ranking.
- 03Early warning on competitor moves inside the LLM layer, before they show up in pipeline as lost deals.
- 04A new category of marketing analytics that didn't exist 18 months ago, in a defensible position before the incumbents catch up.
Role
Designed, built, and shipped end-to-end - product, prompts, and infrastructure.
Status
Live - internal
React
Tap if this resonated.
What do you think?
Leave a comment or ping me — any feedback, thoughts, or collab ideas, I'll really appreciate it. Building alone is no fun.