Skip to Main Content

RASHOMON — AI That Questions Its Own Conclusions What if four cameras witnessed the same event, but none told the complete story? Every day, important decisions are made using incomplete evidence. Security footage misses critical moments, eyewitness accounts contradict each other, and AI systems often produce confident answers even when the evidence is insufficient. The problem isn't always that information is unavailable. It's that we struggle to distinguish what actually happened from what we think happened. RASHOMON is a multi-camera AI investigation system designed to solve this problem. Using four synchronized phone cameras, RASHOMON analyzes an event from multiple perspectives and checks individual claims against the recorded evidence. Rather than generating a single potentially misleading explanation, it classifies each claim as Supported, Contradicted, or Insufficient Evidence. What makes RASHOMON different is its ability to question its own conclusions. Powered by Gemini 4's multimodal capabilities and two Managed Agents, the system uses an Investigator to actively search for missing evidence and a Skeptic to challenge conclusions that might be wrong. Imagine someone claims, "I never moved that object." RASHOMON examines the recordings, identifies relevant moments, and determines whether the evidence contradicts the statement. But if another camera reveals uncertainty, the system can reconsider rather than force a confident answer. Crucially, the AI agents never directly control the verdict. They gather and challenge evidence, while deterministic rules make the final decision. Every conclusion is linked to inspectable footage and a transparent investigation trail. RASHOMON explores applications in security investigations, incident reconstruction, and evidence verification, where premature conclusions can have serious consequences. We're building AI that doesn't just answer questions about reality. It knows when reality hasn't provided enough evidence to answer.

Team

yourselfthere04

More from Gemini 4 NYC Hackathon

We design an agentic system to automatically create + manage intelligent AI opponents/NPCs in video games. The suite of agents analyzes user gameplay…

Adaptive NPCs project preview

Manually routing PCBs is time-consuming and cost-inefficient. Current best algorithms (freerouting, PCBRouter) score very low on PCBWorld's hardest PCB…

routr project preview

Heliocam is a camera app where Gemini controls the camera. It watches what you're pointing at, sets the ISO, shutter, focus, white balance and zoom for you…

heliocam project preview

Repairs and assembly fail because manuals are lost, vague, missing, or nonexistent, so people break things or throw them out. KLIK turns a photo of any object…

Klik project preview

Iron Pulse turns your laptop into a 3D boxing ring. Fight an adaptive AI rival in real time using just your webcam. Throw real punches and lean to dodge…

Iron Pulse Arena project preview

Text your raw footage to an iMessage number and get a finished, trend-informed Reel back in the same thread.

Sticky project preview