Three reasons why we’re pivoting to SkyNet - Jon Auerbach

Three reasons why we’re pivoting to SkyNet

Dear Jarvis team,

It’s hard to believe that just 18 months ago, Jarvis was just a pipe dream drawn on a napkin that was then taped to another napkin because the idea was just so big. What started as Jeff and me working 23 hours a day servicing every apartment in our Reed Street building for free has morphed into a multi-city network of hard-working Jarvi. And we couldn’t be prouder of the work you guys and gals have done, with literally no training other than a couple of YouTube videos we “suggested” you watch. Our members are more organized, more efficient, and just plain happier than they ever could have achieved on their own, and that’s all thanks to you.

So that’s what makes today’s announcement so difficult. After many late nights of contemplation on the beach during our executive off-site in Monterey, we’ve decided to take Jarvis in another direction.  

No, we’re not shutting down.

We’re pivoting.

The reasons are multitude, but we’ve boiled it down to three so the read time doesn’t go above 10 minutes. We’re also changing the name of the company to SkyNet, and effective next week, you will be responsible for building and maintaining the robots that will be taking over your routes.

1. We couldn’t figure out how to scale fast enough

In theory, our plan was always to collect enough data to optimize our Jarvis routes, so that one Jarvis could service 17 different clients in a single day. While we could sign up entire apartment buildings to use our Jarvi, we wanted to give our engineers a challenge, and routing a Jarvis to a bunch of apartments in the same building is so basic. If we had to choose between having the best tech in the industry and making money as a glorified property management company, we would choose tech every single time. It’s not even close.

That being said, the economics of sending a single person to different buildings in different locations really does not work, even with our world-class staff of operations research Ph.Ds fine-tuning routing algorithms 24/7 . At best, we were losing $100/customer/week. During a particularly intense peyote session, it hit us that even if we reached 100,000 subscribers, that would just mean we were losing even more money. Something had to change.

2. We want to capitalize on the machine learning wave

Everywhere you look, there is another story about some startup applying “machine learning” to really stupid applications, like cooking pancakes or making juice.

These ideas are small potatoes and we wanted to think big.

Like change the world big.

And what would change the world?

Robots.

Not just glorified vacuum cleaners, but real actual robots like the ones sci-fi books and movies have been promising us for the past 70 years.

So we spent several nights on the beach hacking together a robot from parts we borrowed from my Tesla. We designed it to carry the leftover artisanal water from our cabins back to our cars as an internal proof of concept to use around the office. Sure, its machine vision mistook shampoo for water and it kept tripping on the uneven wooden floors, but we were astounded that by the end of the retreat, the Jarvis 800’s water recognition ability had improved an astounding 10541%. We knew then that we had stumbled upon something special.

Thanks to a quick funding round we closed on the way back from Monterey via FaceTime, we’re now sitting on a $706 million war chest and a valuation that makes us one of the top five robotics companies in the world. You may have seen the announcement about the machine learning team we poached from Uber. These guys and gal are going to be instrumental in building a production-ready Jarvis 800 in three months, so please give them a hearty Jarvis welcome when they stop by your house to fit you with a camera rig that will observe your every movement and brainwave during your shifts over the next several months.

3. We want to fully implement our culture

When you all signed on to become Jarvi, you received a 154-page book outlining our culture, credos, and what we expect from the people who are privileged enough to wear the Jarvis hooded sweatshirt. And, overall, based on an analysis of your emails and texts, we concluded that 70% of you were indeed living and breathing the Jarvis way in your professional and personal lives. That’s great! But we can do better, and thanks to the Jarvis 800, we will, because each robot will be hard-wired with our culture and therefore will have no choice in the matter.

Doing that presented some really tough programming challenges, and therefore the first version of the Jarvis 800 will only be programmed to follow Jarvis Credo (JC for short) #1: Do not be evil.

Many of you probably wondered why that is item number one in the book, because it’s fairly obvious and didn’t we just copy that directly from Google, and these are good questions.

To answer the second question, we did not copy it from Google. Their credo, which they don’t even use any more, was “Don’t be evil.” That’s such a flippant statement because of its use of the contraction “don’t” that everyone in the Valley was always making fun of them behind their backs. We chose to spell out the whole word to reinforce that at Jarvis, we do not want you to be evil. Being evil is bad and at Jarvis, we’re committed to doing good. So it’s pretty obvious why we would make that the number one most important thing we want to teach our independent contractors and robots.

Now, of course, we recognize that a robot’s view of evil might not be in sync with typical moral norms, and so we’re working as hard as we can to finish hardwiring the remaining credos, before the Jarvis 800 starts dispassionately poisoning clients it considers to be evil. You can be rest assured of that.


We can’t wait to begin the next chapter of SkyNet’s ascendance as the premier on-demand home-management robotics company in the world, and we’re so excited for you to join us on this journey. And if the whole thing ends up causing a nuclear apocalypse, well, we are working on figuring out how to send a reprogrammed Jarvis 800 back in time, so don’t worry.

Sincerely,

John S. Calvin

Chief Robotics Officer, SkyNet


Leave a Reply



^