An AI pitches startup ideas

Take a look at this list of 13 hot startups, from a list compiled by Alex Bresler. Perhaps one of them is the next Juicero?

FAR ATHERA: A CLINICAL AI PLATFORM THAT CAN BE ACCESSED ON DEMAND.

ZAPSY: TRY-AT-HOME SERVICE FOR CONSUMER ELECTRONICS.

MADESS: ON-DEMAND ACCESS TO CLEAN WATER.

DEERG: AI RADIOLOGIST IN A HOME

SPER: THE FASTEST, EASIEST WAY TO BUY A HOME WITHOUT THE USER HAVING TO WEAR ANYTHING.

PLILUO: VENMO FOR B2B SAAS.

LANTR: WE HELP DOCTORS COLLECT 2X MORE ON CANDLES AND KEROSENE.

ABS: WE PROVIDE FULL-SERVICE SUPPORT FOR SUBLIME, VIM, AND EMACS.

INSTABLE DUGIT: GITHUB FOR YOUR LOVED ONES.

CREDITAY: BY REPLACING MECHANICAL PARTS WITH AIR, WE ELIMINATE INSTALLATION COMPLEXITY AND MAINTENANCE HEADACHES, LEADING TO SIGNIFICANT EFFICIENCY GAINS IN PRODUCTION COSTS AND HARVESTING TIME.

CREDITANO: WE BUILD SOFTWARE TO ENABLE HIGH FUNCTIONALITY BIONICS FOR TREATING HEALTH CONDITIONS TO BE SOFTWARE ENGINEERS FOR FREE IN EXCHANGE FOR A FULL STACK SOLUTION THAT DELIVERS WORKFLOW AND EMBEDS ANALYTICS WITHIN THE ELECTRONIC MEDICAL RECORD.

PATT: CONNECTED TOYS THAT ENABLE TWO-WAY VOICE MESSAGING BETWEEN CHILDREN AND THEIR LOVED ONES CAN KNOW THEY ARE GONE, SO THEY GET FREE WATER REFILLS FROM OUR POOL OF VETTED ENGINEERING AND PROGRAMMING FOR KIDS 6-12 WITH A SINGLE MESSAGE.

GITS: WE NOW WANT TO BRING BOXING AS AN ALTERNATIVE TO PAGERS.

As you've probably guessed by now, these are not real startups. (Though given Juicero, you'd be forgiven if you didn't.) Alex Bresler created a generating neural network (GNN) from YCombinator's archive of startup names and descriptions, and the GNN created almost 5000 new ones. (You can download an Excel file with all of them here, to save load on Alex's server.) I picked a few by hand to create the list above, but there was no shortage of funny ones (and plenty of all-too-real ones as well).

Alex used his fundmanageR package (available on Github) to download the YCombinator database into R.  The Generational Neural Network was trained using the keras package for R on these data, using Tensorflow as the back end.  You can find all the details on how the list was generated, including the R code, at Alex's blog linked below.

ASBC LLC: Juice, Err, Ohh…