The problem
Search engine optimisation has thirty years of tooling. Answer-engine optimisation, meaning how often ChatGPT, Gemini or Claude mention your product when someone asks a question in your category, has almost none. GEEKY AIQ gives a marketing team that measurement: which prompts matter, who gets cited, and how the share moves week to week.
How it works
A customer defines the prompts and competitors that matter to them. On a schedule, the platform runs those prompts across models, extracts citations and mentions, and stores the results as a time series. Dashboards show citation share against competitors, and reports highlight what changed since the last scan.
The interesting part
Running the same prompt across several providers on a schedule is exactly the failover, usage-accounting and error-classification problem I worked on inside laravel/ai. One provider rate-limits, another runs out of credit, a third returns a different usage object. Building the scanner on those patterns is what made it tractable for a small team; the SDK work and the product work fed each other.