Explore, Prototype, Learn

How We Get to Good Ideas Faster

We’re right in the middle of our quarterly explore sprints: three day sprints designed to help us quickly express concepts use cases by building prototypes.

And when we say building, we mean it — we’re creating real, working prototypes that people can touch, experience and play with, from concept to build in just three days. We’re doing fast iterations, from Post-it ideas to paper concepts to coded-up systems, to identify which are the right proofs-of-concept to create in longer term projects. In the past couple of weeks, we’ve built everything from an artificial intelligence tool that can combat bias in hiring, to a data platform for biometric data, to an alternative credit score system for renters.

Check out a sampling of the prototypes CoLab teams have created this week:

Tobi: A Hiring AI

Anand Upender, Takashi Wickes, Constantine Pitsilos, Kenny Okagaki

Tobi is an AI created for recruiters that helps them filter candidates. Training from a pool of past makeathon applicant data, the team built a machine learning algorithm that could quickly identify good hires, a chatbot built on Slack to help quickly gauge collaborative ability, a voice-based interview assistant, and a web portal for recruiters to view skill webs for potential candidates. The Tobi platform turns the static process of applying via resumes and cover letters into a dynamic experience a company can have with a large pool of potential hires to identify good fits.

Verismart: Humanizing Smart Contracts

Ethan Ouimet, Sisi Messick, William Wu, Nolan Reis

Verismart is a platform for humanizing smart contracts. Inspired by the recent frenzy of ICOs with complex terms that have the potential to put investor capital at risk, Verismart enables a Wikipedia-like group of “code lawyers” to review the terms coded into smart contracts and translate them into terms a regular person can understand, which platform users could flag as unusual, letting the crowd figure out where the code deviates from the expected.

FonzieKart: Automated Self Repair

Fred Barnhill, Joanne Cheung, Bruno Olmedo, John Oudsteyn

Fonzie is a blockchain-based network that tracks personal automobile usage and leverages data from a network of similar vehicles to provide predictive maintenance. The team built a prototype on MarioKart, linking the the gas pedal to track the amount of acceleration time and decrement the amount of gas in a simulation. Auto-refill mode automatically refills the gas tank when it drops below 20%. Manual-refill allows the player to play until the gas tank drops to 0%. if the tank reaches zero, the player is forced to suspend play for 10 seconds to simulate car maintenance time. This demo allows players to experience the time versus cost tradeoff in car maintenance.

What’s the point of all of this work? We believe that building prototypes is about getting to good ideas faster. And that’s pretty important, because we’re interested in how emerging technologies like blockchain, machine learning, augmented reality and the internet of things are going to change the world. Like Alan Kay said, “The best way to predict the future is to invent it.”