CrashCost

Bunq Hackathon 7.0 — Multimodal AI financial co-pilot for road incidents

In the first 30 minutes after a crash, drivers make decisions that cost them thousands — admitting fault prematurely, missing critical evidence, not knowing whether to claim. CrashCost gives them a real-time verdict: what this will cost, who is likely at fault, and exactly what to do next.

🌍 The Problem In the first 30 minutes after a crash — the most financially consequential window — drivers make decisions that cost them thousands:

Admitting fault prematurely Failing to capture critical evidence Not knowing whether to file a claim or self-pay Accepting incorrect at-fault outcomes The impact is not marginal — it is systemic.

Studies show that 15–25% of disputed insurance claims involve incorrect fault attribution.

This means millions of drivers either:

pay when they shouldn’t lose no-claim bonuses face increased premiums for years All because of what happens in the first few minutes after an incident.

Yet today, drivers receive no real-time financial or legal guidance in this critical moment.

💡 The Solution CrashCost turns bunq into a real-time financial co-pilot during a crash.

Using video, audio, images, and location data, it tells the user:

What this will cost — and what to do next

In under 20 seconds.

🎬 User Scenario James is rear-ended at a roundabout.

Shaken, he opens CrashCost and starts recording.

The app captures video, audio narration, and photos The AI analyzes the full scene in real time While speaking, James says: “I think it was my fault…”

CrashCost immediately intervenes:

🔴 “Do NOT admit fault — evidence suggests the other driver may be responsible.”

The system then:

Identifies missing evidence (“📸 Capture the other vehicle’s license plate”) Estimates repair cost vs insurance excess Predicts financial impact if he files a claim Recommends whether to claim or self-pay Within minutes:

A structured report is generated Financial consequences are clear A bunq payment request is ready to send 🚀 What It Does User records the scene (video + voice narration) and photographs both vehicles AI analyses all evidence and returns a structured incident report in ~20 seconds User settles directly via bunq — payment request or bunq.me link, pre-filled from the AI cost estimate