# Debate brief: Canadian used-car AI MVP The plan under review is `/Users/aniket/Desktop/ArcadeLabs/_canadian_auto_ai/MVP-PLAN.md` (v2). Read it fully first. ## Owner facts (fixed, do not argue these away) - Target: independent used-car lots in Canada (province not yet known). The owner can reach 2–3 such dealers but won't talk to them until there's a demo to show. - Built in-house by a small team. The owner wants a go/no-go in about 3–4 weeks. - These lots probably keep no inventory on a computer. The owner's idea: do an in-person lot audit, then the dealer manages stock in our dashboard (add, edit, mark sold). - Pricing: a monthly fee first, then CPA, or both. - Attribution must be robust, not goodwill. - Unknown: dealer margins, lead volume, whether they have Facebook Pages or ad accounts. ## Gaps the orchestrator spotted (cover any that sit in your lane) - Facebook Marketplace is free, and many independents already sell from it. Why pay for our ads? - Financing is often why buyers pick an independent lot. v2 only hands it off. - Trade-ins. - Competitors: AutoTrader.ca dealer products, dealer website and chat vendors, AI sales-assistant (BDC) tools. - Our own cost to serve one dealer (audit, ads management, transcript review). Does it scale? - How we sell it, and what happens if the dealer won't mark cars sold. ## Rules - The goal is the best possible MVP plan, not politeness. Disagree hard, with reasons. - Numbers: cite a source URL, or label them ASSUMPTION. - You may use web search (mode "standard") for facts. Never put any personal or company names from this workspace into searches. - Write only inside `/Users/aniket/Desktop/ArcadeLabs/_canadian_auto_ai/debate/`. Don't contact anyone. Don't edit MVP-PLAN.md. - Plain, short writing. No filler.