Kite: A search engine for mobile
I’m very proud to share a start-up Jason Hreha and I have been working on since early 2013.
It’s called Kite and it’s a search engine for Android.
Jason and I are proud to announce that we’ve been acquired by Quixey and will be joining their team as Directors of Product. (Techcrunch, Morgan Linton)
Kite is designed to field queries from the user like “best burrito in sf” and return a set of apps that we think can help you accomplish this goal.
The great thing is that if you have the app installed, in many cases, we deeply-link you to the results page for the specific thing you searched for.
If you don’t have the app installed, we send you to the mobile web version of the same content.
The understand the needs of users when a query was entered, we used some natural language processing techniques combined with named entity recognition to understand some of the “things” a user is looking for.
The engine categorizes and classifies apps according to this need, and with some magic from the web crawl and a little bit of magic, we were able to build a graph describing why people use apps, and sameAs relationships between native apps and web apps.
Bartelligence: Bar Management on the Web
I’m very proud of the team at Bartelligence releasing the most recent version of our application.
The project began in 2010 as a simple web inventory tool. I designed a really simple web spreadsheet tool with asynchronous saving and an intuitive interface.
Since then it has grown to manage purchasing and product management. It can send orders directly to vendors, and can analyze price discrepancies and optimize what customers purchase to save them time and money.
Today the team is selling the product to some of the best bars in the country and truly disrupting the space. Congratulations!
A Truly Fantastic Collaboration
Jason Hreha and I have been working on several projects which share the constraint of being able to build and publish within 10 days. One of these I’d like to share is Truly Fantastic.
Fantastic is a platform for supporting the very best artists with small cash donations.
The idea of microdonations was a present in some forms in the early web, but probably could not be executed well enough without very high volume and technologies to minimize transaction costs.
And there are some folks doing a good job on the tipping and gifting front. And of course, assurance contacts incentivize the production of great content and products, but for video content in particular, wouldn’t it be great if viewers of great content could pay only after watching something amazingly awesome?
Our focus was to create a great video experience with curated (and eventually original content) and then buttress this with a highly focused and clean payment flow.
Mockups by Jason Hreha
We like this project. It’s staying up and we’re keeping it alive in our back pocket while we seek out the right artists for the platform.
Cool graphical highlights from Cryptolog, Spring 1997 (recently declassified from Top Secret by NSA)
Many endeavors are a balance of big thinking and vision with small tactics and execution.
The only computer game I’ve played in my adult life is Starcraft 2. SC2 strategy is divided between the macroeconomic management and micromanagement of units in battle.
The speed of which micro is executed by professional players is measured in actions (unit commands) per minute (APM). Professionals can reach the incredible rate of 300-500 APM:
Thankfully, Charles and Ray Eames help us to keep the perspective:
Microwave Motion Sensors for Fun and Profit (HB100 and Arduino)
I read a lot of science fiction and cyberpunk growing up, and I would obsess over certain sci-fi movies, good and bad. One of the latter which I recently have been thinking a lot about is Johnny Mnemonic.
There is a scene in the beginning of the film where he places a device on a door which alerts him to an attack in progress on the other side of the door. I decided to build something similar.
For the microwave sensor, I chose the HB100, it’s just a few dollars and has the range and precision necessary to carry out the task.
Since my processing and logic will be carried out with an Arduino, a simple op-amp circuit is required to amplify the signal to something the arduino can detect. (See page 5 of the design notes.)
After constructing this, I can verify that the signal is being amplified correctly:
The last component necessary to measure precise velocity is a frequency pre-scaler, to divide the frequency of the signal emitted by the HB100 to something in the range of what the Arduino can measure. According to this Frequency Counter library this maxes out at around 8 Mhz. (More information about prescalers: http://www.qsl.net/n9zia/pre/index.html)
At the 10.525 GHz frequency level for my HB100, the rough calculation for velocity is Doppler Frequency = 19.49V (where V is velocity in km/hr).
Putting it all together, it if definitely less aesthetically pleasing than the motion sensor in the movie:
So, let’s output the signal to the serial port. My simple Arduino script:
Then I use the GQ Optics library to graph the signal in a rolling chart:
Rolling motion data! I can detect motion through the wall near my door, and about 30, 40 feet away. Unlike passive infrared, it cannot be fooled by strange heat signatures. And of course, microwaves pass through walls. However, for drone-on-drone warfare, we can definitely expect pretty poor resilience of the microwave signals since it will be quite easy to jam/mask.
Weekly Inspiration: Flowing
I’m a huge fan of Roxy Music and Bryan Ferry. There is some sort of glamorous desperation in many of their best songs. The 70’s is full of transitions. One of my favorite songs is Mother of Pearl; about 20 seconds into it, the song transitions from a raucous party to a sober and reflective ballad.
Oh looking for love
In a looking glass world
Is pretty hard for you
Your own predilection
The search for perfection
Goes on and on and on and on
You’re highbrow, holy
With lots of soul