The Problem
Buying a car is designed to disadvantage you. Dealer inventory is opaque, pricing is theatrical, and the negotiation process rewards people who’ve done it dozens of times over first-time buyers. The information asymmetry is the whole business model.
I decided to close the gap with AI.
The Approach
Phase 1: Market Research
- Used CarEdge (caредge.com) to pull real dealer inventory data, market-value pricing, and days-on-lot statistics for target vehicles in a 200-mile radius
- Fed that data into Claude to identify specific VINs where price-to-market-value spread suggested negotiating room
- Built a monitoring script that scraped CarEdge daily and alerted when target vehicles appeared or prices dropped
Phase 2: Negotiation Prep
- Claude synthesized a negotiation playbook based on the specific vehicle, dealer history, and market conditions
- Prepared counter-offer sequences, out-the-door price calculations (including all dealer fees), and objection responses
- Identified 3 comparable vehicles at competing dealers to use as leverage
Phase 3: The Purchase
- Entered negotiations with printed CarEdge market data and specific VIN comparisons
- Used the playbook sequence: start with the out-the-door number, not the monthly payment
- Closed $3,200 below sticker without a trade-in
What Made It Work
The CarEdge “days on lot” metric was the secret weapon. A vehicle that’s been sitting for 60+ days costs the dealer ~$1,200/month in floor plan financing. That’s real leverage. Knowing a specific VIN had been there 73 days changed the entire dynamic of the conversation.
The monitoring script meant I wasn’t relying on memory or manual checking — I was notified the day a target vehicle hit 60 days on lot.
The Playbook
The full process is documented in a reusable template:
- Define target vehicle specs + acceptable color/option combinations
- Pull CarEdge market data for zip codes within radius
- Build VIN shortlist sorted by days-on-lot descending
- Research dealer history and incentive timing (end of month/quarter)
- Generate negotiation script with specific numbers and fallback positions
- Execute with printed data in hand
Total research time: 6 hours spread over 2 weeks. Outcome: $3,200 saved vs. asking price.