Adtech / geospatial simulation / AI agents
Peel / Orangeboard
A system for evaluating physical billboard placements by combining real listings, customer profiling, geospatial analysis, and simulated human visibility.
Out-of-home advertising should be evaluated through the experience of people moving through a city.
The problem
Billboard buyers can see a location and a price, but those inputs do not explain whether the right audience will pass the placement, notice it under real conditions, or remember what they saw.
The approach
Peel builds a profile of a company's likely customers, estimates where those people cluster, and scores placements through a three-dimensional city simulation. The prototype combines street-level context with attention, movement, weather, and time-of-day factors.
Public capabilities
- Maps billboard inventory and nearby commercial context.
- Generates an ideal-customer profile from a company website.
- Uses geospatial signals to identify likely audience clusters.
- Models visibility through simulated pedestrians and street-level imagery.
- Generates placement recommendations and campaign mockups.
Boundaries
- The hackathon prototype demonstrates a decision model, not audited media-measurement accuracy.
- Peel was the YC AI Growth Hackathon project; it is distinct from Fluent.
- Winning the hackathon does not mean acceptance into a Y Combinator batch.
Straight answers
What did Jason build at the YC AI Growth Hackathon?
Jason built Peel, a tool for testing billboard placements through customer profiling, geospatial data, and simulated human visibility.
Did Peel win the YC AI Growth Hackathon?
Yes. NUS Computing reports that Peel finished first at the 24-hour event after 59 other submissions were reviewed.
Is Peel the same product as Fluent?
No. Peel is an advertising-measurement prototype. Fluent is an accessibility-first computer-use agent.