How Did One Simple Internal Email From the CEO Propel a Struggling Company to Become a $1 Trillion USD Corporation and a Leader in AI?

Companies embracing digital transformation (DX) and Artificial Intelligence are experiencing amazing results. Structured data makes AI work. How can your company catch up? Find here the three basic ways.

In 2002 Jeff Bezos issued a corporation-wide mandate with a simple email. It went more or less like this:

TO: All Development

SUBJECT: Bezos Mandate

All teams will henceforth expose their data and functionality through service interfaces. Teams must communicate with each other through these interfaces.

There will be no other form of inter-process communication allowed: no direct linking, no direct reads of another team's data store, no shared-memory model, no back-doors whatsoever. The only communication allowed is via service interface calls over the network.

It doesn't matter what technology they use.

All service interfaces, without exception, must be designed from the ground up to be externalizable. That is to say, the team must plan and design to be able to expose the interface to developers in the outside world. No exceptions.

Anyone who doesn't do this will be fired.

Thank you; have a nice day!

Jeff Bezos

Yes, Mr. Bezos saw the power of structured data and the benefits of data communication through APIs. Of course, that was just the beginning. From that point forward, Amazon took off to become a trillion-dollar company, and the giant we all now know. Bezos transformed his retail business. He took his company from a struggling traditional organization model that was having serious trouble supporting its own growth (too many products, too much volume, business processes were highly inefficient because the primitive software was breaking down under pressure, etc.), and far from being profitable, to a business resting on an integrated, highly modular digital foundation...a business that was fully digital and eventually powered by AI. Of course we also know that during the transformation process, one of the most powerful cloud services was born - AWS. There is a lot to learn from them.

The questions you might be asking yourself right now are:

• What is structured data?
• Is my corporation, in the industry we operate in, suitable for digital transformation?
• If so, how can I transform my company to be fully digital?

WHAT IS STRUCTURED DATA?

Your company data is generated in various ways: text, voice, video, images, spreadsheets, etc. A small percentage of that data might perhaps be structured, but the bulk of it is not. Let’s illustrate it this way: imagine the Rubik’s cube. If you got yours brand new, all the rows and columns on each face had only one color before it was scrambled. in a perfect world, this is how your data set should look- very well arranged and presented in such a way that it is ready for communication through the interface.

Why did I use the Rubik's cube as an example? Because each face is composed of many little units. Depending on how many columns and rows there are, you can classify the cubes as 2x2, 3x3, 4x4, etc. (and yes, there are some silly 1x1 cubes out there). But each little unit is still there and is part of each of the 6 sides of the cube. With data, I like to imagine each little unit is an attribute. The most important element of your data, albeit the smallest, is the attribute. This raises the question: "What are data attributes?" Simply put, data attributes are the characteristics of an object. From a data science view, they are the features of a data set. It applies to all data across all industries. For example, let's say you are an online retailer and have among all your SKUs a blue striped shirt. Well, the objective attributes are blue, shirt, size, stripe, brand, etc. However, there are also equally important subjective attributes such as: formal shirt, informal shirt, dinner shirt, wedding shirt, beach shirt, royal blue, turquoise blue, baby blue, sky blue, etc. Of course, attributes could be more complex, but this is a simple way to describe them.

I can almost guarantee that most of your company data is not well structured. Going back to the Rubik’s cube analogy, your corporate data looks like a group of scrambled Rubik’s cubes tossed around in the corner of a room. Each Rubik cube represents a database, file cabinet, hard drive of an office computer, data in the cloud, etc. Not a pretty picture.

All those scrambled Rubik's cubes now must be arranged and will have to be put together or converted into a large one, with many units per side. I will call it the platform. In the beginning it will look like a 20x20 Rubik’s cube or perhaps larger, but as your company grows it will have no problem expanding and growing along with you to become a 30x30, 100x100, 500x500, etc. Your platform will allow you to grow seamlessly anywhere in the world.

Structured data is the fuel for AI. So, if you are eagerly trying to adopt AI for some projects you have in mind, you will soon find out your data might pull you back. It needs to be structured first. Typically this is an arduous, manual job, but it can also be automated. I will discuss this subject another time.

A Brief History of the Multinational Corporations and Why Your Data is Not Structured

The first public multinational corporation was registered around 1602: The Dutch East India Company (VOC). They have been, so far, the richest company the world has ever seen, and the most valuable company in history. Yes, converted to today’s value it would be worth over $9 Trillion USD. 

If you added together Apple, Amazon, Alphabet and Microsoft, still they would not be worth as much as The Dutch East India Company. In fact, they would barely account for a little more than half of it.

To manage over 150 merchant ships and 50,000+ employees in many countries, The Dutch East India Company had to give their managers a lot of latitude. They behaved as autonomous units. Remember, phone lines and email did not exist. So, silos of information were formed to operate efficiently given the circumstances. Flexible management and siloed operating architecture started back then!

As time passed, during the 17th and 18th century, financial services and trading companies became a little more sophisticated and creative in how they ran their business operations. Manufacturing lagged behind because it was mostly an artisan process done primarily in private homes. The information generated by all corporations was preserved across multiple internal units, however, and not unified. The 19th century saw the industrial revolution and manufacturing took off as an important driver of the world economy. Mass production was invented, and manufacturing facilities increased in size considerably. They were no longer household affairs. Yet, the way all corporations operated and shared information internally and externally, did not change. Moving forward, we reach the 20th century with all of its inventions. 

The most plausible invention, perhaps, was the computer and software. Corporations adopted ERPs and all kinds of enterprise software systems. But these systems did not change how corporations operate. They simply took what was already in place and just made it more efficient. Systems adapted to the way companies generate data, in silos.

Now in the 21st century we are facing an unprecedented time in history. Operating models can be fully digital. We can use AI to support corporate decisions. However, AI runs only on structured data, hence the importance of structured data and the creation of a corporate digital platform. The AI platform is the decision engine powering the digital operating model of the 21st century corporation!

Can you see now the wisdom behind the “Bezos mandate”? Amazon became a trillion-dollar company. Had they not transformed to digital, I would venture to say, they would have been long gone. The operating model set by The Dutch East India Company has been finally broken for good.

Corporate decisions are now made by software powered by AI. Entire processes that were performed by humans have now been digitized. For instance, bank loans are now approved by AI. Prices at some of your favorite online stores have been left to the wisdom of AI. Corporations can forecast the demand of their products thanks to AI. Financial companies in the futures market can now predict the yield of crops for the year with satellite images, weather data, etc. Motor racing sports can improve vehicle design and help drivers to improve their times with the help of AI processing all their biometric data. Process manufacturing companies can predict the quality of their output product from the early stages of the process, avoiding waste and saving energy, thanks to AI. And the list goes on and on. Decision making has been industrialized. Predictions are generated by strong analytic algorithms guiding a variety of operations within the walls of a corporation. Algorithms manage the flow of information. In other cases, they guide how products should be built, handled, and delivered. At the very center of the digital corporation is the AI platform, with structured data feeding AI project after AI project. The cycle, with a few exceptions, is very simple:

User engagement (or sensors in an industrial production environment) allows for data collection. The data, thereafter, propels algorithm design. The models give birth to predictions and, after the performance is evaluated, proper improvements proceed. Humans are pushed to the edge of this process: They will make long term decisions (management), and perform other manual activities the AI platform reliably recommends in order to continue delivering value (work force).

WHICH INDUSTRIES AND COMPANIES CAN ADOPT ARTIFICIAL INTELLIGENCE AND AN AI-DRIVEN OPERATION PROCESS?

The simple answer- All companies can take advantage of AI.

Early adopters will have a tremendous advantage against competitors. Just in the United States alone, the following companies using AI to make day to day decisions have taken off, leaving competitors behind: Amazon, Uber, Netflix, Google and Peloton, just to mention a few. The list can go on and on. You can see how they are all different and belong to different industries. Yet, they have one thing in common. They have fully digitized their processes and fully embraced AI. Had these companies not adopted AI, they would probably be unknown to most of us, if they even existed at all. Some of them were born with digital processes at their core, but most of them had to experience what can sometimes be painful- transformation.

Yes, the adoption of AI is not without any pains or hiccups. The biggest change involves the culture of the company. The decision to transform has to come from the top, cascading down to the bottom of the pyramid, and not the other way around.

Companies can no longer afford to keep data in silos, where it helps no one or perhaps only benefits a small group of people. A digital platform of structured data ready for AI projects is needed. The change cannot be done overnight, but once started can influx a new life to your company...a much-needed infusion of new energy, so to speak. The CEOs who realized this are in a frenzy to undergo digital transformation and AI adoption.

All of this seems very clear, logical and simple, right? But here is the tricky part, and where most companies struggle right now: what labor force is needed to successfully transform a company to digital? What kind of team do you need to gather together and empower to start the digital revolution inside your company? How many data scientists, statisticians, computer engineers, front end engineers, back end engineers, code developers, mathematicians, test engineers, etc. should you hire, or reassign from other internal tasks? What kind of computing infrastructure do you need to develop this platform? There is no clear answer here that fits every company in every situation because it will depend on the scope of the project, the industry you are in, and what use cases will become priorities. Will it be sales, manufacturing, finances, marketing? It makes sense to start with a project that will have the most impact with the least amount of investment required.

Developing an AI platform will take time. Amazon took at least 6 years to go from digital transformation to AI. However, as mentioned before, and it is no secret, you must start with structured data.

There are ways you can enter the AI superhighway and deploy a solution relatively fast by structuring your data quickly. If you are serious about bringing the power of AI to your corporation, contact me through a private message and I will be happy to guide you. 😊