Take-Home Authorship Probe — live-interview challenge packs for hiring leads at founder-led SaaS

A pre-final-round gate for hiring leads at founder-led SaaS companies. The hiring manager pastes a candidate's take-home submission (code, design exercise, analytical write-up) and the role's expected competencies. The product runs an adversarial multi-model debate against the submission and returns a 30-minute live interrogation pack: 8-12 targeted questions plus one injected modification the candidate must extend on the spot. The pack is engineered to expose whether the candidate genuinely authored and understood their own work — separating practitioners from AI-augmented submitters who can't defend their submission under live probing. AE's adversarial multi-model debate generates probes a single interviewer wouldn't think of, and the 6-pattern autopsy taxonomy (Cosmetic Confidence, Concession Laundering, Fatal Grounding Immunity) maps directly onto the failure modes that distinguish authored from AI-augmented work. Each pack is graded post-interview by which questions actually surfaced gaps, building a per-role library of high-yield probes the hiring lead reuses.

Take-homes are broken by LLMs, hiring managers are already buying defenses, and AE's adversarial-debate plus autopsy-taxonomy stack maps directly onto the exposure problem — sold async to a solo-operator-friendly buyer at price points that hit the ARR target.

Killed: evidence_search_exhausted.