01
The stories
Every story is pulled straight from a target firm's own newsroom announcement. Nothing goes up before 1 July 2026, and every source is cited and linked at the bottom, so you can go and read the original yourself.
How it works
02
The question
Each interview prompt is written by hand alongside the story. That's deliberate: a question written by someone who actually read the deal announcement holds up better under a follow-up than one assembled from a headline.
03
The grading
Every question carries a model answer framework, hand-built for that specific prompt, setting out the reasoning an interviewer would actually expect to hear. Grading runs against a fixed rubric of defined categories, structure, commercial link, legal relevance, and the strength of your own view. Claude applies that rubric and produces a score against it. It's explicitly instructed to credit a differently reasoned but still valid argument rather than penalise you for not matching the framework word for word, and every response is parsed and validated before it reaches you.
04
The tracking
Every attempt is logged against its topic. Your dashboard shows a real average and best score per area, and exactly where you need another rep.
05
The matching
Firm matching runs on a weighted scoring algorithm built from your answers to 15 questions. Each answer maps onto measurable traits, scored against how every firm actually looks on practice strength and culture. The same input produces the same result every time.