WHY IT MATTERS: Digital Transformation

Amazing insights into the mobile industry state and trends.

WHY THIS IS IMPORTANT

Andreessen Horowitz, and Benedict Evans in particular, have been known to spot trends long before everyone else. Here again, it appears they are making predictions about retailing - following in the steps of the print industry when Internet came along. Not certain I agree about the parallel but I agree that mobile and AI and other technologies will impact and transform the retail world in the near future. However, as opposed to the print media, retailers often sell physical goods that cannot, for now, be 3D printed at home. So either you pick them up in the store or get it delivered by FedEx or a drone. 

That being said, retail stores are bound to transform in showrooms and service centers as I wrote about in the past. And this is a major shift that many will fail to take.

Take a minute to watch the video. This is the future of retailing.

WHY THIS IS IMPORTANT

Amazon has disrupted many industries with its online business. Recently, as other digital leaders have done, Amazon is trying to bring digital revolution in the world to blend our physical and digital experiences - think Google self driving cars, Starbuck mobile phone payment and geofencing app, Apple stores that are more showrooms and help centres than stores because eCommerce is there to fulfill orders to your door, etc.

Amazon has been focussed on grocery for the past 10 years to use this weekly recurring, low cost, high volume shopping trip to the supermarket as its pathway to bring digital into the real world and blend it with online e-commerce.

With Amazon Fresh for examples they have been busy putting trucks on the road to deliver groceries to your door overnight with minimal costs. Why grocery? Because its high volume guarantees that trucks will be on the road everyday in all areas. And once trucks are running around carrying groceries they can also bring other items - books, TVs, etc. - at little or no cost, effectively making FedEx irrelevant or at least giving Amazon huge negotiation power.

I wrote about this in many blog posts, just see here: http://www.scoop.it/t/digital-transformation-of-businesses?q=amazon 

Here with Amazon GO, it addresses some of the major pain points of customers in grocery stores: waiting in line at the cash register. As they've done online, Amazon focusses on its customer issues to smash them or make them so irrelevant that they are not issues anymore. See "amazon secret sauce" for insights on this: http://sco.lt/979c4P

Take Amazon Prime for example. It addresses the delivery fee issue: we do not want to have to pay for delivery. So they make it a non issue, providing 2-day delivery for free for an annual fee. Side effect: if you've paid for prime, then you spend up to 5x more at Amazon (300$ annually for average client, 1500$ for prime member), basically becoming a loyal shopper. Wow.

Back to Amazon GO and why it is important. The grocery industry has not seen a major disruption since Costco started top sell grocery in bulk format years ago - grocers ost 10% market share with that new player in the field. But today, grocery stores are the same, the layout is the same, the process is the same than it was 50 years ago. There has been little or no innovation in stores recently (last one may be the barcode in 1970s) but GO shows that there may be a perfect storm of technologies that combine to make physical stores transformation a reality:

1- mobile phones are on every customer back pocket;

2- electronic payment is everywhere, making cash almost irrelevant;

3- loyalty programs and big data allows retailers unprecedented customer knowledge and forecasting trends;

4- artificial intelligence enables real-time image analysis, feature detection, and robotic improvements.

From the concept store video and the Amazon description, all that GO is doing is applying these new technologies on a new problem set: adding and removing items to shopping basket and speeding the checkout process.

Of course, at the end of the day, the winner will not be the grocer with the best technology. It will be the one with the lowest prices and the best selection (best marketing and customer service won't hurt either). On those fronts (price & selection) Amazon GO faces huge headwinds from established grocers that have long standing relationships with suppliers and way more volume than Amazon. But it does not mean Amazon cannot provide customers with a more enjoyable, easy, fast experience - especially to millennials - which may give GO 5% or 10% of the market (the "good" 5% by the way, those that spend more and spend on higher margin items) from established retailers. And this will hurt established grocers like hell.

Also see the analysis from business insider here: http://www.businessinsider.com/amazon-go-grocery-store-future-photos-video-2016-12/ 

and the plans for Amazon to open up to 2000 store in the future if GO concept works: http://www.businessinsider.com/amazon-2000-grocery-stores-10-years-2016-10

If you have kids, have them read this.

Very technical series of articles on deep learning coding techniques. Useful to read even if you have only limited DL coding experience because it pulls the covers from over a very new way of coding - especially for old nerds like me!

part 1: http://online.cambridgecoding.com/notebooks/cca_admin/deep-learning-for-complete-beginners-recognising-handwritten-digits 

part 2: http://online.cambridgecoding.com/notebooks/cca_admin/convolutional-neural-networks-with-keras 

part 3: http://online.cambridgecoding.com/notebooks/cca_admin/neural-networks-tuning-techniques 

A 3 part post that explains what CNNs are. It explains their architecture, and provides plenty of examples and diagrams to help even non technical readers understand the concepts.

WHY THIS IS IMPORTANT

CNNs are essential technologies behind the current boom in artificial intelligence and deep learning that has led us to have very good speech recognition in Google Android phones, amazing image recognition, self driving cars, etc. etc. They are essential components of any digital transformation strategy.

A reference chart of neural network architecture, with detailed explanation of each one. Very useful reference.

Read the article but you can skip the 7min video. For a better - and more relaxed - discussion, consider watching this 1hr interview full of descriptions of self-driving car insights: https://www.youtube.com/watch?v=2zy_07g2IrM

For a more technical speech, view the Berkley presentation, even though there are elements that are redundant it provides additional insights into the technology: https://www.youtube.com/watch?v=Hxoke1lDJ9w

WHY THIS IS IMPORTANT

His approach is radically different than others and more "purist" from an artificial intelligence and crowd sourcing and I believe that makes his approach something to consider in other areas.

For more reading on the subject, this bloomberg paper also brings good insights: http://www.bloomberg.com/features/2015-george-hotz-self-driving-car/

Moreover it appears that Google lead is slowly shrinking: http://www.bloomberg.com/news/articles/2016-09-12/google-car-project-loses-leaders-and-advantage-as-rivals-gain 

Fatal car crash of a Tesla on highway using automated driving mode. The post provides some information about the crash, which appears to be due to a condition that was never programmed into the system (side hit).

WHY THIS IS IMPORTANT

Autonomous vehicles will become more and more popular in the future and this event reminds us of the complexities involved. We are very early in this field and much like aviation there should be formal investigations and corrective actions taken when such crash occur so that all manufacturers can learn from crashes. This calls for an international database of car crashes and there should be mandatory requirements for new self-driving systems to "pass" the exam ie. to ensure they all react correctly to previously seen car crashes and other incidents. 

This thus becomes an opportunity to learn from the global past experience of millions of drivers and billions of kilometers driven in all conditions, across all countries. I am not aware of such a central database but would expect it to become a reality if we want autonomous driving to be a transformation beyond what we have today: each driver learning for themselve.

Image aesthetics is difficult to quantify adn thus difficult for computers to do: how can a digital system determine what makes a photograph "good"? This paper describes how one company has used deep learning techniques to do so.

WHY THIS IS IMPORTANT

With results like this one, we must all question what is possible and open our strategies to include things that in the past have proven difficult or impossible to do. To be successful, strategist must however ensure that they have at their disposal the raw materials required to train computer systems. In the past this would not have been as important but today's new algorithm require massive amounts of data and the resources to train machine intelligence in a good way.

For s shorter, less technical version of this article, go to https://medium.com/stories-from-eyeem/how-we-trained-an-algorithm-to-predict-what-makes-a-beautiful-photo-e8de8bccd642#.f6u7gxxwx 

218 slides of jammed packed information. Soon I will be able to blog one slide of this Mary Meeker State of the Internet annual presentation per day for the whole year. Wow...

WHY THIS IS IMPORTANT

This is an annual event for digital experts: the state of the Internet. This year, it focusses on deep learning (think voice+image recognition) and cars. Because really the Internet growth is slowing down. So is mobile phone usage. The trend thus appears to be what we build on top of the Internet and not the Internet itself. Finally.

Every company executive and board member should listen to this talk. Even if you don't understand everything you should ask yourself: how is our company positioned to address those trends? In most cases I assume the answer will be: we are not. This should be cause for concern and action.

For previous year presentations, go to http://www.scoop.it/t/digital-transformation-of-businesses/?q=meeker