This article titled "Wie Unternehmen vom Vormarsch des maschinellen Lernens profitieren können" appeared in German last week in the "Digitaliserung" column of Wirtschaftwoche.

When a technology has its breakthrough, can often only be determined in hindsight. In the case of artificial intelligence (AI) and machine learning (ML), this is different. ML is that part of AI that describes rules and recognizes patterns from large amounts of data in order to predict future data. Both concepts are virtually omnipresent and at the top of most buzzword rankings.

Personally, I think – and this is clearly linked to the rise of AI and ML – that there has never been a better time than today to develop smart applications and use them. Why? Because three things are coming together. First: Users across the globe are capturing data digitally, whether this is in the physical world through sensors or GPS, or online through click stream data. As a result, there is a critical mass of data available. Secondly, there is enough affordable computing capacity in the cloud for companies and organizations, no matter what their size, to use intelligent applications. And thirdly, an "algorithmic revolution" has taken place, meaning it is now possible to train trillions of algorithms simultaneously, making the whole machine learning process much faster. This has allowed for more research, which has resulted in reaching the "critical mass" in knowledge that is needed to kick off an exponential growth in the development of new algorithms and architectures.

We may have come a relatively long way with AI, but the progress came quietly. After all, during the last 50 years, AI and ML were fields that had only been accessible to an exclusive circle of researchers and scientists. That is now changing, as packages of AI and ML services, frameworks and tools are today available to all sorts of companies and organizations, including those that don't have dedicated research groups in this field. The management consultants at McKinsey expect that the global market for AI-based services, software and hardware will grow annually by 15-25% and reach a volume of around USD 130 billion in 2025. A number of start-ups are using AI algorithms for all things imaginable – searching for tumors in medical images, helping people learn foreign languages, or automating claims handling at insurance companies. At the same time, entirely new categories of applications are being created whereby a natural conversation between man and machine is taking center-stage.

Progress through machine learning

Is the hype surrounding AI and ML even justified? Definitely, because they offer business and society fascinating possibilities. With the help of digitization and high-performance computers, we are able to replicate human intelligence in some areas, such as computer vision, and even surpass the intelligence of humans. We are creating very diverse algorithms for a wide range of application areas and turning these individual pieces into services so that ML is available for everyone. Packaged into applications and business models, ML can make our life more pleasant or safer. Take autonomous driving: 90% of car accidents in the US can be traced to "human failure". The assumption is that the number of accidents will decline over the long term if vehicles drive autonomously. In aviation, this has already been reality for a long time.

MIT pioneers Erik Brynjolfsson and Andrew McAfee predict that the macroeconomic effect of the so-called "second machine age" will be comparable to what the steam engine once unleashed when it replaced humans' muscular strength ("the first machine age"). Many are uncomfortable with the idea that an artificial intelligence exists alongside human intelligence. That is understandable. We must therefore discuss – parallel to the technological developments – how humans and AI can co-exist in the future; the moral and ethical aspects that arise; how to ensure we have a good grip on AI; and which legal parameters we need in order to manage all this. Answering these questions will be just as important as the effort to solve the technological challenges, and neither dogmas nor ideologies will help. Instead, what's needed is an objective, broad-based debate that takes into account the wellbeing of society as a whole.

Machine Leaning at Amazon

For the past 20 years, thousands of software engineers at Amazon have been working on ML. We dare to claim that we are the company that has been applying AI and ML as a business technology the longest. We know that innovative technologies always take off whenever barriers to entry fall for market participants.

That is the case right now with AI and ML. In the past, anyone who wanted to use AI for himself had to start from scratch: develop algorithms and feed them with enormous amounts of data – even if he later needed an application for a strictly confined context. This is referred to as so-called "weak" AI. Many of the consumer interfaces that everyone is familiar with today, such as recommendations, similarities or autofill functions for search prediction – they are all ML driven. In the meantime, they can predict inventory levels or vendor lead times, detect customer problems and automatically deduct how to solve them; and discover counterfeit goods and sort out abusive reviews, thereby protecting our customers from fraud. But that is only the tip of the iceberg. At Amazon, we are sitting on billions of historical order information data, which allows us to create other AI/ML-based models based on AI for many different kinds of functionalities. For example programming interfaces that developers can use to analyze images, change text into true-to-life language or create chatbots. But ultimately, there is something to be found for everyone who wants to define models, train them, and then scale. Pre-configured, attuned libraries and deep learning frameworks are widely available, which allow anyone to get started very fast.

Companies like Netflix, Nvidia, or Pinterest use our capabilities in ML and deep learning. More and more layers are being created in a kind of ecosystem on which companies and organizations can 'dock' their business – depending on how deep they want to, and are able to, immerse themselves in the subject matter. Decisive is the openness of the layers and the reliable availability of the infrastructure. In the past, AI technologies were so expensive that it was hardly worth it to use them. Today, AI and ML technologies are available off the shelf, and they can be called up according to one's individual requirements. They form the basis for new business models. Even users who are not AI specialists can very easily and affordably incorporate the building blocks into their own services. In particular small and medium-sized companies with innovative strength can benefit. They do not have to learn any complex ML algorithms and technologies, and they can experiment without incurring high costs.

Artificial intelligence helps to satisfy the customer

One of the most advanced areas of application is e-commerce. AI-supported pre-selection mechanisms help companies to free their customers' decision making from complexity. The ultimate goal is customer satisfaction. If there are only three types of toothpaste, the customer can easily pick one and feel good about it. When more than 50 kinds are on offer, the choice becomes complicated. You have to decide, but you're not sure if the decision is the right one. The more possibilities there are, the more difficult it becomes for the customer. Our best-known algorithms come from this field: filtering product suggestions based on one's purchase history of products with similar attributes, or on the behavior of other customers who were interested in similar things.

Of course, consistent quality also contributes to the satisfaction of the customer. Intelligent support makes life easier for the provider and the customer. For Amazon Fresh, for example, we have developed algorithms that learn how fresh groceries have to look, how long this state lasts, and when food should no longer be sold. Airlines or rail transport companies could also use this for their quality control by running an algorithm based on the image data of the freight; the algorithm would recognize damaged goods and automatically sort them out.

If you can predict demand, you can plan more efficiently

In B2B and B2C businesses, it is critical that goods are available quickly. It is for this reason that we at Amazon have developed algorithms that can predict the daily demand of goods. This is particularly complex for fashion goods, which are always available in many different sizes and variations and for which reorder possibilities are very limited. Information about past demand, among others, is fed into our system, as well as fluctuations that can occur with seasonal goods, the effect of special offers, and the sensitivity of customers to price shifts. Today we can predict precisely how many shirts in a certain size and color will be sold on a defined day. We have tackled this issue and made the technology available to other companies as a web service. MyTaxi, for example, benefits from our ML-based service to plan at what time and at which place the customer will need the vehicle.

New division of labor

But AI is much more than just forecasting. In the field of fulfillment, which is relevant for numerous industry sectors, we are thinking of ideas of how AI can contribute the most to taking another step away from a Tayloristic work pattern. Applied in robots, AI can free people from routine activities that are physically difficult and often stressful. Machines are very good at, and sometimes even outperform, tasks that are complicated for a human to do, such as finding the optimal route in a warehouse for a certain number of orders and transporting heavy goods to the point where it is sent to the customer. For supposedly easy tasks, by contrast, the robot is overwhelmed; an example is recognizing a box that has landed on the wrong shelf. So how to bring together the best of both players? By letting intelligent robots learn from humans how to identify the right goods, take on various orders and navigate their way autonomously through the warehouse on the most efficient route. This is how we take away the most tedious part of the task and shift resources to more interaction with the customer.

Our client SCDM uses the core idea of freeing up resources for "human" strengths, but in a completely different context. SCDM is a service provider that supports banks and insurance companies with digitization. Using AI, SCDM enables its customers to classify documents that are of very different formats (PDF, Excel or photos), for example a report about the performance of an investment product that contains hundreds of pages. By scanning hundreds of thousands of documents simultaneously, SCDM's algorithm recognizes which document is relevant for a specific request, finds out where relevant data for a specific type of preparation is located, and then extracts the data from the document. As a result, there is less bias and fewer errors in the number crunching, and more time for human interaction with important stakeholders like investors, analysts and other customers.

Machine learning in education, medicine and development aid

In addition to their potential for things like efficiency and productivity, ML and AI can also be used in education. Duolingo, which offers free language course apps, uses text-to-language algorithms to assess and correct learners' pronunciation. In medicine, AI supports doctors in analyzing X-Ray CTs or MRT images. The World Bank also uses AI in order to implement infrastructure programs, development aid and other measures in a more targeted manner in the future.

More room for optimism

Despite all these developments, many people from academia, business and government have a critical view of ML and AI. There have been warnings that a new super-intelligence is jeopardizing our civilization – and these warnings have been effective in attracting publicity.

However, neither hysteria nor euphoria should be allowed to get the upper hand in the public debate. What we need instead is a pragmatic-optimistic view of the emerging possibilities. AI enables us to get rid of tasks in our work which damage our health or where machines are better than we are. Not with the goal of making ourselves redundant. Rather, in order to gain more personal and economic freedom – for interpersonal relationships, for our creativity and for everything that we humans can do better than machines. That is what we should strive for. If we don't, we will ultimately forego the economic and societal opportunities that we could have grasped.

Improving Customer Service with Amazon Connect and Amazon Lex

| Comments ()

Customer service is central to the overall customer experience that all consumers are familiar with when communicating with companies. That experience is often tested when we need to ask for help or have a question to be answered. Unfortunately, we've become accustomed to providing the same information multiple times, waiting on hold, and generally spending a lot more time than we expected to resolve our issue when we call customer service.

When you call for customer assistance, you often need to wait for an agent to become available after navigating a set of menus. This means that you're going to wait on hold regardless of whether your issue is simple or complex. Once connected, the systems that power call centers generally don't do a good job of using and sharing available information. Therefore, you often start out anonymous and can't be recognized until you've gone through a scripted set of questions. If your issue is complex, you may end up repeating the same information to each person you talk to, because context is not provided with the handoff. It's easy to end up frustrated by the experience, even if your issue is successfully resolved.

At Amazon, customer obsession is a fundamental principle of how we operate, and it drives the investments we make. Making sure that customers have a great experience when they need to call us is something that we've invested a lot of time in. So much so, that in March 2017, we announced Amazon Connect, which is the result of nearly ten years of work to build cloud-based contact centers at scale to power customer service for more than 50 Amazon teams and subsidiaries, including, Zappos, and Audible. The service allows any business to deliver better over-the-phone customer service at lower cost.

When we set out to build Amazon Connect, we thought deeply about how artificial intelligence could be applied to improve the customer experience. AI has incredible potential in this area. Today, AWS customers are using the cloud to better serve their customers in many different ways. For instance, Zillow trains and retrains 7.5 million models every day to provide highly specific home value estimates to better inform buyers and sellers. KRY is helping doctors virtually visit patients and accurately diagnose aliments by applying machine learning to symptoms. Netflix is using machine learning to provide highly personalized recommendations to over 100 million subscribers. There are really exciting projects everywhere you look, including call centers.

When Amazon Connect launched, we spoke about the integration with Amazon Lex. One of the really interesting trends in machine learning lately has been the rise of chatbots, because they are well suited to fulfilling customer requests with natural language. Amazon Lex, which uses the same conversational technology as Amazon Alexa, is Amazon Web Services' deep-learning powered chatbot platform. By linking Amazon Lex chatbots into the Amazon Connect contact flow, customers are able to get help immediately without relying on menus or specific voice commands. For example, an Amazon Lex driven conversation with your dentist's office might look like this…

Connect: "Hello, thanks for calling. Is this Jeff?"

Jeff: "Yes"

Connect: "I see you have a cleaning appointment this Friday. Are you calling to confirm?"

Jeff: "No, actually."

Connect: "Ok, what are you calling about?"

Jeff: "I'd like to change my appointment to be next Monday."

Connect: "No problem, I have availability on Monday July 3rd at 11:00 AM. Does that work?

Jeff: "That's fine."

Connect: "Great. I have booked an appointment for you on Monday, July 3rd at 11:00 AM. Is there anything else I can help you with?

Jeff: "Can you send me a text confirmation?"

Connect: "Sure. I have sent a text message confirmation of your appointment to your cell. Can I do anything more for you?"

Jeff: "No, that's great. Bye."

The chatbot is able to quickly and naturally handle the request without waiting for an agent to become available, and the customer was never presented with menus or asked for information the office already had. AWS Lambda functions made the corresponding calls to the database and scheduling software, making sure that the interaction happened quickly and at extremely low cost. The workflow-based functionality of Amazon Lex and Amazon Connect also helps to reduce mistakes by making sure interactions play out consistently every time.

If the customer's issue is not able to be resolved by the chatbot, Amazon Lex is able to pass on the full context of the conversation to a human representative. This keeps the customer from wasting time repeating answers to questions and lets the representative focus 100% of their time on solving the problem, which increases the odds the customer is going complete the call feeling positive about the experience.

Today, we're announcing the general availability of Amazon Lex integration with Amazon Connect. We've also enhanced the speech recognition models used by Amazon Lex to support integration with other call center providers as well, so that all telephony systems can start using AI to improve customer interactions.

We think artificial intelligence has a lot of potential to improve the experience of both customers and service operations. Customers can get to a resolution fast with more personalization, and human representatives will be able to spend more time resolving customer questions.

Getting Started: Amazon Connect is available to all customers in the US East (N. Virginia) region. You can get started by visiting Additional information on Amazon Lex integration can be found at

Stop waiting for perfection and learn from your mistakes

| Comments ()

This article titled "Wartet nicht auf Perfektion – lernt aus euren Fehlern!" appeared in German last week in the "Digitaliserung" column of Wirtschaftwoche.

"Man errs as long as he doth strive." Goethe, the German prince of poets, knew that already more than 200 years ago. His words still ring true today, but with a crucial difference: Striving alone is not enough. You have to strive faster than the rest. And while there's nothing wrong with striving for perfection, in today's digital world you can no longer wait until your products are near perfection before offering them to your customers. If so, you will fall behind in your market.

So if you can't wait for perfection, what should you do instead? I believe the answer is to experiment aggressively with your product development, accepting the possibility that some of your experiments will fail.

Anyone who has listened to, or worked with, management gurus know their mantra: Failure is a necessary part of progress. That's true, but there's often a big gap between the management theory and the reality on the ground. People want to experiment and learn from things that go wrong. But in the flurry of day-to-day business, they're not given enough time to really reflect on the cause of an error and what to do differently next time.

The solution is to find a systematic approach that prevents errors from repeating themselves.

From perfection to anti-fragility

In finding such a systematic way, you first need to distinguish between two types of errors that can happen in your company: those of technology and those of human decision-making. The nice thing is: if you know how to deal effectively with the first, you might end up being better in the second, making better decisions. The financial mathematician and essayist Nassim Taleb offers an interesting take on this issue. He has argued that errors are incredibly valuable because they lead to innovation. He uses the term 'anti-fragility' to make his point. Today's digital business models require smaller, frequent releases to reduce risk. That means the technologies underpinning these new business models must be more than just robust. They must be 'anti-fragile'. The main feature of anti-fragile technology is that it can 'err' without falling apart. In fact, a crisis can make it even better.

At Amazon, we also require our systems and customer solutions to be anti-fragile, and we do that by designing our systems to stand the test of time. Our systems must be able to evolve and become more resilient to failure. They must become more powerful and more feature-rich over time as a result of learning from customer feedback and any failure modes they may encounter while operating the systems.

An example of a German company that has become 'anti-fragile' is HARTING, the world's leading provider of heavy pluggable connectors for machines and plants. HARTING shows how to think a step ahead about the meaning of quality standards in the digital world. Quality and trust are the most important values for this traditional company, and Industry 4.0 and the digital transformation have already been important focus areas for them since 2011. Even though it was hard to accept at first, HARTING has meanwhile realized that errors are inevitable. For that reason, its development switched to agile methods. It also uses the "minimum-viable-product" approach and relies on microservices for its software. Working this way, HARTING can discard things and create new things more easily. All in all, HARTING has become faster.

That can be seen with HARTING MICA, an edge computing solution that enables older machines and plants to get a digital retrofit. The body and hardware still reflect HARTING's standard of perfection. But for the software, the goal is "good enough", because a microservice is neither ever finished nor perfect. As a result, wrong decisions and mistakes can be corrected very quickly and systems can mature faster, approaching the state of antifragility. If the requirements change or better software technologies become available, each microservice can be thrown out and a new one created. That's how you gain speed and quickly digitize old machines and connect them to the cloud within a manageable cost framework.

Taking the dread out of mistakes

If you want to become anti-fragile, more than robust, like HARTING and other companies, you need to proactively look for the weak spots in a system as you experiment. In a system that should evolve, all sorts of errors will happen that you weren't able to predict, especially when systems need to scale into unknown territories. So subject your system to continuous failures and make subsystems artificially fail using tools like Netflix's Chaos Monkey.

If you do all of this, you will start to objectify errors at your company and make dealing with errors a matter of normality. And when errors become 'business as usual', no one will be afraid of taking a risk, trying out a new idea, a new product or a new service and seeing what happens when customers interact with it. That's how you quickly find solutions that really work in the future.

At Amazon, our approach for systematically and constructively dealing with errors is called the "cause of error" method. It refrains from seeking "culprits". Instead it documents learning experiences and derives actions that ultimately improve the availability of our systems.

From root cause to innovation

The method first calls for fixing an error by analyzing its immediate root cause and taking steps to mitigate the damage and restore the initial running state as quickly as possible.But we are not content with that result. We go further, trying to extract the maximum amount of insight from the incident. And this process begins as soon everything is working again for the client.

A key element of our cause-of-error method is asking 5 'Why?' questions (a technique that originated in quality control in manufacturing). This is important because it determines the fundamental root of the problem.

Take the case of a website: Why was it down last Friday? The web servers reported timeouts. Why were there timeouts? Because our web services are overloaded and couldn't cope with the high traffic. Why were the web servers overloaded? Because we don't have enough web servers to handle all requests at peak times. Why don't we have enough web servers? Because we didn't consider possible peaks in demand in our planning. Why didn't we take peaks in demand into account in our planning? By the end of this process, we know exactly what happened and which clients were affected. Then we're in a position to distill an action plan that ensures that specific error doesn't happen again.

Quite often, applying this cause-of-error approach allows us to find breakthrough innovations, in the spirit of Nassim Taleb. That's how the solution Auto Scaling was created, after a certain client segment was fighting with strongly fluctuating hits on their website. When the load increases for a website, Auto Scaling automatically spins up an additional web server to service the rising number of requests. Conversely, when the load subsides, Auto Scaling turns off web servers that are not needed in order to save cost.

What it reveals is: Organizations need to look beyond superficial success. This is true for the development of systems as well as business models. If you want to remain agile in a complex environment, you must follow this path, even if it means leaving the comfort zone. If we transfer these ideas into an organizational context, three aspects might be worth considering:

1. Embrace error as a matter of fact

Jeff Bezos once said about Amazon: "I believe we are the best place in the world to fail." That inspires a lot of our people to experiment, find errors and turn them into something innovative. A statement like this encourages your people to actively look for errors, and to turn them into pieces of innovation. And: reward employees when they find errors. What we have learned from our development work at Amazon is that you need to always look beyond the surface of an error. Some of our best products have been born from errors.

2. Make due with incomplete information

German companies have a tradition of being thorough and perfectionist. In the digital world, however, you need to loosen those principles a bit. Technology is changing so fast; you need to be fast too. Make decisions even if the information you have is not as complete as you would like.Jeff Bezos put his finger on that when he wrote in his most recent letter to shareholders that "most decisions should probably be made with somewhere around 70% of the information you wish you had. If you wait for 90%, in most cases, you're probably being slow. Plus, either way, you need to be good at quickly recognizing and correcting bad decisions. If you're good at course correcting, being wrong may be less costly than you think, whereas being slow is going to be expensive for sure."

3. Praise the value of learning

I've stressed the need for companies to have a systematic approach to how they deal with errors. But your approach will only work if it's part of your overall culture. Make sure you understand your DNAandknow what people are thinking and talking about on the work floor. Openly praising experimentation in product development and encouraging people to find errors will come across as empty rhetoric if your employees really do have reason to fear repercussions for themselves personally if they make mistakes.

It is a matter of leadership to foster and shape a culture of experimentation that is practiced day in, day out.

Whatever companies come up with in order to systematically learn from mistakes, it will make them better in competing in the digital world. And it will give them the freedom and courage to take their systems, solutions and business models to a higher level.

Today, I'm excited to announce the general availability of Amazon DynamoDB Accelerator (DAX), a fully managed, highly available, in-memory cache that can speed up DynamoDB response times from milliseconds to microseconds, even at millions of requests per second. You can add DAX to your existing DynamoDB applications with just a few clicks in the AWS Management Console – no application rewrites required.

DynamoDB has come a long way in the 5 years since we announced its availability in January 2012. As we said at the time, DynamoDB was a result of 15 years of learning in the area of large scale non-relational databases and cloud services. Based on this experience and learning, we built DynamoDB to be a fast, highly scalable NoSQL database to meet the needs of Internet-scale applications.

DynamoDB was the first service at AWS to use SSD storage. Development of DynamoDB was guided by the core set of distributed systems principles outlined in the Dynamo paper, resulting in an ultra-scalable and highly reliable database system. DynamoDB delivers predictable performance and single digit millisecond latencies for reads and writes to your application, whether you're just getting started and want to perform hundreds of reads or writes per second in dev and test, or you're operating at scale in production performing millions of reads and writes per second.

Saving crucial milliseconds

Having been closely involved in the design and development of DynamoDB over the years, I find it gratifying to see DynamoDB being used by more than 100,000 customers - including the likes of AirBnB, Amazon, Expedia, Lyft, Redfin, and Supercell. It delivers predictable performance, consistently in the single-digit milliseconds, to users of some of the largest, most popular, iconic applications in use today. I've had a chance to interact with many of these customers on the design of their apps. These interactions allow me to understand their emerging needs, which I take back to our development teams to further iterate on our services. Many of these customers have apps with near real-time requirements for accessing data that need even faster performance than single-digit milliseconds. These are the apps that have motivated us to develop DAX.

To give you some examples of my interactions, I've been talking to a few ad-tech companies lately, and their conversations are about how they can save milliseconds of performance. For their applications, they have 20-50 ms to decide whether or not to place a bid for an ad. Every millisecond that is spent querying a database and waiting for a key piece of data is time that they could otherwise use to make better decisions, process more data, or improve calculations to place a more accurate bid.

These high-throughput, low-latency requirements need caching, not as a consideration, but as a best practice. Caches reduce latencies to microseconds, increases throughput, and in many cases, help customers save money by reducing the amount of resources they have to overprovision for their databases.

Caching is not a new concept, and I have always wondered, why doesn't everyone cache?

I think the reasons are many, but most follow a similar trend. Although many developers are aware of the patterns and benefits of adding a cache to an application, it's not easy to implement such functionality correctly. It's also time consuming and costly. When you write an application, you might not need or design for caching on day one. Thus, caching has to be shoehorned into an application that is already operational and experiencing load that would necessitate the added benefits. Adding caching when your app is already experiencing load is not easy. As a result, we see many folks trying to squeeze out every last drop of performance, or significantly overprovision their database resources to avoid adding a cache.

Fully managed cache for DynamoDB

What if you could seamlessly add caching to your application without requiring a re-write?

Enter DynamoDB Accelerator. With the launch of DAX, you now get microsecond access to data that lives in DynamoDB. DAX is an in-memory cache in front of DynamoDB and has the identical API as DynamoDB. There's no need to rewrite your applications to access your cache. You just point your existing application at the DAX endpoint, and as a read-through/write-through cache, DAX seamlessly handles caching for you. Microsecond response times, millions of requests per second—and of course, it's a fully managed environment that is highly available over multiple Availability Zones so you no longer have to worry about managing your cache.

With DAX, we've created a fully managed caching service that is API-compatible with DynamoDB. What this means to you as a developer is that you don't have to re-write your DynamoDB application to use DAX. Instead, using the DAX SDK for Java, you just point your existing application at a DAX endpoint, and DAX handles the rest. As a read-through/write-through cache, DAX will intercept both reads and writes to DynamoDB. For read-through caching, when a read is issued to DAX, it will first check to see if that item is in cache. If it is, DAX returns the value with response times in microseconds. If the item is not in cache, DAX automatically fetches the item from DynamoDB, caches the result for subsequent reads, and returns the value to the application. This is done transparently to the developer. Similarly, for writes, DAX first writes the value to DynamoDB, cache the value in DAX, and then returns success to the application. This way, reads after writes are available for cache hits, which further simplifies the application. With cache eviction handled by time-to-live (TTL) and write-through evictions, you no longer need the code to perform this task. DAX provides all the benefits of a cache, with a much simpler developer experience.

The following is code for an application that talks to DynamoDB:

All you have to do is point your application at the DAX endpoint with three lines of code. You've added in-memory caching without performing brain surgery on the application.

Adding DAX is as simple as the following code:

Why doesn't everyone cache? Many times, it is too costly in terms of time and complexity because developers have to alter some of their most critical code paths. With DAX, you get faster reads, more throughput, and cost savings - without having to write any new code.

What's not to like? This is a fantastic addition for our DynamoDB customers. To get started with DAX today, see Amazon DynamoDB Accelerator (DAX).

Many of our customers share my excitement:

10 billion matches later, Tinder has changed the way people meet around the world. "For Tinder, performance is absolutely key. We are major users of DynamoDB. We love its simplicity and ability to scale with consistent performance," said Maria Zhang, VP of Engineering at Tinder. "With DAX, AWS has taken performance to a new level, with response times in microseconds. We really like how DAX integrates seamlessly with DynamoDB, is API-compatible, and doesn't require us to write any new code. We are excited for the General Availability of DAX."

Careem is a car-booking service and app that serves more than 40 cities and 11 countries in the broader Middle East. The company uses a number of AWS services, including Amazon DynamoDB to store locations of its captains, promotions, and configurations. "We have been involved early on during the DAX public preview, and have been running our production workload on DAX with no issues," said Tafseer-ul-Islam Siddiqui, Software Architect at Careem. "We are using DAX to scale our reads across our network of services. As a write-through cache, DAX has simplified our application stack and has removed the need for building a central service for our caching needs. A key feature that motivated our adoption of DAX was that it is API-compatible with DynamoDB and thus required minimal changes to use with our existing app - you only need to change the DynamoDB client to the DAX client. Our team was really impressed with the built-in failover and replication support."

Canon INC. Office Imaging Products Development Planning & Management Center provides mission-critical cloud services connecting to business machines for worldwide customers across four continents. "Amazon DynamoDB Accelerator (DAX) is a very wonderful service to improve the user experience of Amazon DynamoDB," said Takashi Yagita, Principal Engineer, Office Imaging Products Development Planning & Management Center, Canon INC. "Our developers like the excellent design concept of DAX SDK, which enables us to switch from DynamoDB and start using DAX seamlessly. Our team has succeeded in keeping the DynamoDB capacity units far lower while improving the data access speed by DAX. We welcome that DAX is generally available."

This is a really a fantastic addition for our DynamoDB customers. To get started with DAX today, please see

Expanding the Cloud – An AWS Region is coming to Hong Kong

| Comments ()

Today, I am very excited to announce our plans to open a new AWS Region in Hong Kong! The new region will give Hong Kong-based businesses, government organizations, non-profits, and global companies with customers in Hong Kong, the ability to leverage AWS technologies from data centers in Hong Kong. The new AWS Asia Pacific (Hong Kong) Region will have three Availability Zones and be ready for customers for use in 2018.

Over the past decade, we have seen tremendous growth at AWS. As a result, we have opened 43 Availability Zones across 16 AWS Regions worldwide. Last year, we opened new regions in Korea, India, the US, Canada, and the UK. Throughout the next year, we will see another eight zones come online, across three AWS Regions (France, China, and Sweden). However, we do not plan to slow down and we are not stopping there. We are actively working to open new regions in the locations where our customers need them most.

In Asia Pacific, we have been constantly expanding our footprint. In 2010, we opened our first AWS Region in Singapore and since then have opened additional regions: Japan, Australia, China, Korea, and India. After the launch of the AWS APAC (Hong Kong) Region, there will be 19 Availability Zones in Asia Pacific for customers to build flexible, scalable, secure, and highly available applications.

As well as AWS Regions, we also have 21 AWS Edge Network Locations in Asia Pacific. This enables customers to serve content to their end users with low latency, giving them the best application experience. This continued investment in Asia Pacific has led to strong growth as many customers across the region move to AWS.

Organizations in Hong Kong have been increasingly moving their mission-critical applications to AWS. This has led us to steadily increase our investment in Hong Kong to serve our growing base of enterprise, public sector, and startup customers.

In 2008, AWS opened a point of presence (PoP) in Hong Kong to enable customers to serve content to their end users with low latency. Since then, AWS has added two more PoPs in Hong Kong, the latest in 2016. In 2013, AWS opened an office in Hong Kong. Today we have local teams in Hong Kong to help customers of all sizes as they move to AWS, including account managers, solutions architects, business developers, partner managers, professional services consultants, technology evangelists, start-up community developers, and more.

Some of the most successful startups in the world—including 8 Securities, 9GAG, and GoAnimate—are already using AWS to deliver highly reliable, scalable, and secure applications to customers.

9GAG is a Hong Kong-based company responsible for, one of the top traffic websites in the world. It's an entertainment website where users can post content or "memes" that they find amusing and share them across social media networks. 9GAG generates millions of Facebook shares and likes per month, attracts over 78 million global unique visitors, and receives more than 1 billion page views per month. 9GAG has a small team of nine people, including three engineers to support the business, and uses AWS to service their global visitors.

GoAnimate is a Hong Kong-based company that allows companies and individuals to tell great visual stories via its online animation platform. GoAnimate uses many AWS services, including Amazon Polly, to allow users to make their visual animations speak. They chose to use AWS in order to focus on developing their platform, instead of managing infrastructure. They believe that they have reduced development time from 20 to 30 percent by having done so.

Some of the largest, and most well respected, enterprises in Hong Kong are also using AWS to power their businesses, enabling them to be more agile and responsive to their customers. These companies include Cathay Pacific, CLSA, HSBC, Gibson Innovations, Kerry Logistics, Ocean Park, Next Digital, and TownGas.

Hong Kong's largest listed multimedia group, Next Digital, operates businesses spanning Hong Kong, Taiwan, Japan, and the United States. They operate in an industry where malicious groups frequently launch distributed denial-of-service (DDoS) attacks to disrupt availability. Then too, Internet service providers can shut down their services any time they feel threatened by the DDoS attacks. Next Digital operates on AWS in a more highly available and fault-tolerant environment than their previous colocation solution. Beyond running their web properties and applications, Next Digital also uses Amazon RDS (database), Amazon ElastiCache (caching), and Amazon Redshift (data warehousing). Further, taking advantage of the local AWS Hong Kong-based team, Next Digital uses AWS Enterprise Support for Infrastructure Event Management and other high-touch support services.

Kerry Logistics, a global logistics company based in Hong Kong, runs a number of corporate IT applications on AWS, including its Infor Sun Accounting Environment and Kewill Freight Forwarding Systems across multiple regions on AWS globally. Their goal has been to ensure that their IT infrastructure sits as closely to their customers and users as possible.

In addition to established enterprises, government organizations, and rapidly growing startups, AWS also has a vibrant ecosystem in Hong Kong, including partners that have built cloud practices and innovative technology solutions on AWS. AWS Partner Network (APN) Consulting Partners in Hong Kong help customers migrate to the cloud. APN Consulting Partners include global partners such as Accenture, Datapipe, Deloitte, Infosys, KPMG and Rackspace, and local partners such as ICG, eCloudValley, Masterson, and Nextlink, among many others.

The new AWS Asia Pacific (Hong Kong) Region, coupled with the existing AWS Regions in Singapore, Tokyo, Sydney, Beijing, Seoul, and Mumbai, and a future one in Ningxia, will provide customers with quick, low-latency access to websites, mobile applications, games, SaaS applications, big data analysis, Internet of Things (IoT) applications, and more. I'm excited to see the new and innovative use cases coming from our customers in Hong Kong and across Asia Pacific, all enabled by AWS.

Unlocking the Value of Device Data with AWS Greengrass.

| Comments ()

Unlocking the value of data is a primary goal that AWS helps our customers to pursue. In recent years, an explosion of intelligent devices have created oceans of new data across many industries. We have seen that such devices can benefit greatly from the elastic resources of the cloud. This is because data gets more valuable when it can be processed together with other data.

At the same time, it can be valuable to process some data right at the source where it is generated. Some applications – medical equipment, industrial machinery, and building automation are just a few – can't rely exclusively on the cloud for control, and require some form of local storage and execution. Such applications are often mission-critical: safeties must operate reliably, even if connectivity drops. Some applications may also rely on timely decisions: when maneuvering heavy machinery, an absolute minimum of latency is critical. Some use cases have privacy or regulatory constraints: medical data might need to be stored on site at a hospital for years even if also stored in the cloud. When you can't address scenarios such as these, the value of data you don't process is lost.

As it turns out, there are three broad reasons that local data processing is important, in addition to cloud-based processing. At AWS we refer to these broad reasons as "laws" because we expect them to hold even as technology improves:

  1. Law of Physics. Customers want to build applications that make the most interactive and critical decisions locally, such as safety-critical control. This is determined by basic laws of physics: it takes time to send data to the cloud, and networks don't have 100% availability. Customers in physically remote environments, such as mining and agriculture, are more affected by these issues.

  2. Law of Economics. In many industries, data production has grown more quickly than bandwidth, and much of this data is low value. Local aggregation and filtering of data allows customers to send only high-value data to the cloud for storage and analysis.

  3. Law of the Land. In some industries, customers have regulatory or compliance requirements to isolate or duplicate data in particular locations. Some governments impose data sovereignty restrictions on where data may be stored and processed.

Today, we are announcing the general availability of AWS Greengrass, a new service that helps unlock the value of data from devices that are subject to the three laws described above.

AWS Greengrass extends AWS onto your devices, so they can act locally on the data they generate while still taking advantage of the cloud. AWS Greengrass takes advantage of your devices' onboard capabilities, and extends them to the cloud for management, updates, and elastic compute and storage.

AWS Greengrass provides the following features:

  • Local execution of AWS Lambda functions written in Python 2.7 and deployed down from the cloud.
  • Local device shadows to maintain state for the stateless functions, including sync and conflict resolution.
  • Local messaging between functions and peripherals on the device that hosts AWS Greengrass core, and also between the core and other local devices that use the AWS IoT Device SDK.
  • Security of communication between the AWS Greengrass group and the cloud. AWS Greengrass uses the same certificate-based mutual authentication that AWS IoT uses. Local communication within an AWS Greengrass group is also secured by using a unique private CA for every group.

Before AWS Greengrass, device builders often had to choose between the low latency of local execution, and the flexibility, scale, and ease of the cloud. AWS Greengrass removes that trade-off—manufacturers and OEMs can now build solutions that use the cloud for management, analytics, and durable storage, while keeping critical functionality on-device or nearby.

AWS Greengrass makes it easier for customers to build systems of devices (including heterogeneous devices) that work together with the AWS Cloud. Our goal is not to provide an alternative for the cloud, but to provide tools for customers to use the cloud to build applications and systems that can't be moved entirely to the cloud. Using AWS Greengrass for local execution, customers can identify the most valuable data to process, analyze, and store in the cloud.

With AWS Greengrass, we can begin to extend AWS into customer systems—from small devices to racks of servers—in a way that makes it easy to do the things locally that are best done locally, and to amplify those workloads with the cloud.

Getting started: AWS Greengrass is available today to all customers, in US East (N. Virginia) and US West (Oregon). You can get started by visiting

In many high-throughput OLTP style applications, the database plays a crucial role in achieving scale, reliability, high-performance, and cost efficiency. For a long time, these requirements were almost exclusively served by commercial, proprietary databases. Soon after the launch of the AWS Relational Database Service (RDS) customers gave us feedback that they would love to migrate to RDS. Yet, what they desired more, was a reality that unshackled them from the high-cost, punitive licensing schemes, which came with proprietary databases.

They would love to migrate to an open-source style database like MySQL or PostgreSQL, if such a database could meet the enterprise-grade reliability and performance these high-scale applications required.

We decided to use our inventive powers to design and build a new database engine that would give database systems such as MySQL and PostgreSQL reliability and performance at scale. Meaning, at a level that could serve even the most demanding OLTP applications. It gave us the opportunity to invent a new database architecture that would address to needs of modern cloud-scale applications, departing from the traditional approaches that had their roots in databases of the nineties. That database engine is now known as "Amazon Aurora" and launched in 2014 for MySQL, and in 2016 for PostgreSQL.

Amazon Aurora has become the fastest-growing service in the history of AWS and frequently is the target of migration from on-premise proprietary databases.

In a paper published this week at SIGMOD'17, the Amazon Aurora team presents the design considerations for the new database engine and how they addressed them. From the abstract:

Amazon Aurora is a relational database service for OLTP workloads offered as part of Amazon Web Services (AWS). In this paper, we describe the architecture of Aurora and the design considerations leading to that architecture. We believe the central constraint in high throughput data processing has moved from compute and storage to the network. Aurora brings a novel architecture to the relational database to address this constraint, most notably by pushing redo processing to a multi-tenant scaleout storage service, purpose-built for Aurora. We describe how doing so not only reduces network traffic, but also allows for fast crash recovery, failovers to replicas without loss of data, and fault-tolerant, self-healing storage. We then describe how Aurora achieves consensus on durable state across numerous storage nodes using an efficient asynchronous scheme, avoiding expensive and chatty recovery protocols. Finally, having operated Aurora as a production service for over 18 months, we share lessons we have learned from our customers on what modern cloud applications expect from their database tier.

I hope you will enjoy this weekend's reading, as it contain many gems about modern database design.

"Amazon Aurora: Design Considerations for HighThroughput Cloud-Native Relational Databases", Alexandre Verbitski, Anurag Gupta, Debanjan Saha, Murali Brahmadesam, Kamal Gupta, Raman Mittal, Sailesh Krishnamurthy, Sandor Maurice, Tengiz Kharatishvili, Xiaofeng Bao, in SIGMOD '17 Proceedings of the 2017 ACM International Conference on Management of Data, Pages 1041-1052 May 14 – 19, 2017, Chicago, IL, USA.

This article titled "Wie die Digitalisierung Wertschöpfung neu definiert" appeared in German last week in the "Größer, höher, weiter (bigger, higher, further)" column of Wirtschaftwoche.

Germany's "hidden champions" – family-owned companies, engineering companies, specialists – are unique in the world. They stand for quality, reliability and a high degree of know-how in manufacturing. Hidden champions play a significant role in the German economy; as a result, Germany has become one of the few countries in Western Europe where manufacturing accounts for more than 20% of GDP. By contrast, neighboring countries have seen a continuous decline in their manufacturing base. What's more, digital technologies and business models that are focused on Industry 4.0 (i.e., the term invented in Germany to refer to the digitalization of production) have the potential to reinforce Germany's lead even more. According to estimates by Bitkom, the German IT industry association, and the Fraunhofer Institute of Industrial Engineering IAO, Germany's hidden champions will contribute a substantial portion to the country's economic growth by 2025 and create new jobs. At the same time, many experts believe the fundamental potential of Industry 4.0 has not even been fully leveraged yet.

The power of persistence versus the speed of adjustment

Most of Germany's hidden champions have earned their reputation through hard work: they have been optimizing their processes over decades. They have invested the time to perfect their processes and develop high-quality products for their customers. This has paid off – and continues to do so.

However, digital technologies are now ushering in a paradigm change in value creation. Manufacturing can be fully digitalized to become part of a connected "Internet of Things" (IoT), controlled via the cloud. And control is not the only change: IoT creates many new data streams that, through cloud analytics, provide companies with much deeper insight into their operations and customer engagement. This is forcing Mittelstand companies to break down silos between departments, think beyond their traditional activities, and develop new business models.

In fact, almost every industrial company in Germany already has a digitalization project in place. Most of these companies are extracting additional efficiency gains in their production by using digital technologies. Other companies have established start-ups for certain activities, or pilot projects aimed at creating showcases. But many of these initiatives never get beyond that point. The core business, which is doing well, remains untouched by all this. And one of the main reasons why is because the people with the necessary IT expertise in Mittelstand companies sometimes are not sitting at the strategy table as often as they should.

Will these initiatives be enough to secure the pole position for Germany's Mittelstand? Probably not. Companies in growth markets are catching up. China's industry, for example, is making huge progress – something that took years to achieve in other places. The role of Chinese manufacturers in the worldwide market is changing: from low-cost workbench to global provider of advanced technology. Market leaders from Germany therefore realize they cannot afford to rest on their laurels.

Competitors from the software side are also reshuffling the balance of power, because their offerings will create a completely new market alongside the traditional business of Mittelstand toolmakers and mechanical and systems engineering companies. If new and innovative companies, such as providers of data analytics, specialized software providers or companies that can bundle complementary offerings, appear on the scene, traditional manufacturing would suddenly become just one module among many – namely manufacturing-as-a-service.

Creating added value in an Industry-4.0 environment often happens when B2B companies integrate B2C approaches, in turn sparking change in their own industry. This requires using agile development processes for continuous improvement and creating a broader portfolio of solutions, for example by increasingly connecting the shop floor with data "from the outside" such as logistics and inventory management. Software that plays an ever-greater role in the "digital factory" of the future will continuously expand its functionalities. Already today, traditional components used in automated industry and made by companies such as Beckhoff, Harting, WAGO, etc. can connect seamlessly to the cloud. Hidden champions from the field of automation technology digitalize their products, enabling their customers to easily join the "smart factory", an environment in which manufacturing facilities and logistics systems are interconnected without the need for a person to operate them. A great example of this kind of digital transformation outside of Germany is General Electric. It is best captured in the words of their CEO Jeffrey Immelt: "If_you went to bed last night as an industrial company, you're going to wake up today as a software and analytics company_."

Efficient individualization

The example of Stölzle Oberglas, a leading Austrian glass producer, shows how an industrial company is able to weave the laws of the consumer industry into its own industry. If a customer decides at the last moment (for example due to a large upcoming sports event) to sell a special edition with the name of the winning team on it, Stölzle needs to deliver at short notice. In the past, this would have been cost-prohibitive to do, but in today's digital age, such a highly customized product must not cost more than an off-the-shelf product. Stölzle can afford that because, with the help of software provider Actyx, it has consolidated data from its entire production process, can analyze the data intelligently, and makes it available for the user. In this way, changing specifications can flow into the production process practically in real time using cloud technology. But client-driven innovation in an Industry 4.0 environment doesn't stop here. Actyx uses the insights gained in this project, continues to develop the solution based on those insights, and makes it available to a broader group of users through its solution portfolio. It is similar to what we do at Amazon Web Services too: we develop new features and services based on concrete feedback from customers and then make them available to all our users.

Ecosystem of additional services Knowing how to connect the knowledge of digital native in a meaningful way with engineering knowledge will be critical for hidden champions' future success. Almost daily, new start-ups are formed in Silicon Valley, Tel Aviv, London and Berlin. The business model of many of these new firms is about creating even greater added value for the user of a machine or device: Using sensors that connect machines and products in the "Internet of Things, other services can now also be created that are no longer limited to the assembly line. At the same time, this kind of experimentation poses only a small risk, because in the cloud, services and the exact server capacity can be reserved for each individual application purpose and then paid per use.

These kinds of services are developed by the Berlin-based company WATTx, an independent spin-off from the 100-year-old heating engineer Viessmann. WATTx was created by the company owners to supplement Viessmann's standard products with intelligent digital services, such as an IoT platform for commercial buildings. According to data from sensors inside and outside the building, heater grids, lighting and window shades can be managed remotely. In the meantime, WATTx is doing much more than that. It brings together all of its digital talents in Berlin and gives them unlimited access to new technologies, such as the cloud. On the one hand, this allows ideas to be realized very quickly –or thrown out quickly if they are not achievable. Ideas are also developed and tested here before they hit the market as new companies. In the meantime, Viessmann is developing services on its own that offer added value related to its basic products of heating and thermostats. By working in this way, this traditional German company is able to maintain the contact to end-customers and explore completely new markets.

Keeping an eye on the big changes

Software and services are areas where a producer of a machine or device initially does not feel at home, simply because software and services were never part of its core business. Changing processes that already work seems to be a high risk, at least in the short term. But if the strategic dimension is lacking in Industry 4.0 projects, many companies may not generate any innovative added value at all — neither at the micro- nor macroeconomic level. In the long term, there is a high risk that more agile competitors will take the lead over 'traditional' industrial companies if the latter fail to develop a new path through the global ecosystem of machines, products and (digital) services. But those industrial companies that do take the bold step of implementing new approaches and solutions by embracing cloud technologies will maintain their hard-won status in the German economy. And there's a good chance they will play a more important role in the future.

Coming to STATION F: The first Mentor's Office powered by AWS!

| Comments ()

I am excited to announce that AWS is opening its first Mentor's Office at STATION F in Paris! The Mentor's Office is a workplace exclusively dedicated to meetings between AWS experts and the startups. STATION F is the world's biggest startup campus. With this special offer starting at the end of June, at the campus opening, AWS increases the support already available to startup customers in France.

All year long, AWS experts will deliver technical and business assistance to startups based on campus. AWS Solutions Architects will meet startup members for face-to-face sessions, to share guidance on how cloud services can be used for their specific use cases, workloads, or applications. Startup members will also have the possibility to meet with AWS business experts such as account managers, business developers, and consultants. They can explore the possibilities of the AWS Cloud and take advantage of our IT experience and business expertise. With these 1:1 meetings, AWS delivers mentoring to startups to help them bring their ideas to life and accelerate their business using the cloud.

AWS will also provide startups with all of the benefits of the AWS Activate program, including AWS credits, training, technical support, and a special startup community forum to help them successfully build their business. For more details about the Mentor's Office at STATION F, feel free to contact the AWS STATION F team.

With this opening, Amazon continues to build out global programs to support startup growth and to speed up innovation. Startups can also apply to other Amazon programs to boost their businesses, such as:

  • Amazon Launchpad, which makes it easy for startups to launch, market, and distribute their products to hundreds of millions of Amazon customers across the globe.
  • Alexa Fund, which provides up to $100 million in venture capital funding to fuel voice technology innovation.

After the launch of AWS in 2006, we saw an acceleration of French startups adopting the cloud. Successful French startups already using AWS to grow their businesses, across Europe and around the world, include Captain Dash, Dashlane, Botify, Sketchfab,Predicsis, Yomoni, BidMotion, Teads, FrontApp, Iconosquare, and many others. They all get benefits from AWS's highly flexible, scalable, and secure global platform. AWS eliminates the undifferentiated heavy lifting of managing underlying infrastructure and provides elastic, pay-as-you-go IT resources.

We have also seen start-ups in France using AWS to grow and become household names in their market segment, such as Aldebaran Robotics (SoftBank Robotics Europe). This startup uses AWS to develop new technologies. They are able to concentrate their engineering resources on innovation, rather than maintaining technology infrastructure, which is leading to the development of autonomous and programmable humanoid robots.

Cloud is also an opportunity for startups to reach security standards that were not accessible before. For example, PayPlug is an online payment by credit card solution enabling e-merchants to enrich the customer experience by reinventing the payment experience. Such a service requires suppliers to get PCI DSS certification for the "Service Provider" level, a very demanding certification level. Using AWS's PCI DSS Level 1 compliant infrastructure, Payplug has been certified by L'ACPR (L'Autorité de contrôle prudentiel et de resolution, the French supervisory for prudential and resolution authority) as a financial institution, a major step in their development.

I look forward to meeting the builders of tomorrow at STATION F in the near future.

Go French Startups!

I will be returning this weekend to the US from a very successful AWS Summit in Sydney, so I have ample time to read during travels. This weekend however I would like to take a break from reading historical computer science material, to catch up on another technology I find fascinating, that of functional Magnetic Resonace Imaging, aka fMRI.

fMRI is a functional imagine technology, meaning that it just records the state of the brain at one particular point in time, but the changing state over a period of time. The basic technology records brain activity by measuring changes in blood flow through the brain. The technology relies on the fact that cerebral blood flow and neuronal activation are coupled. When an area of the brain is in use, blood flow to that region also increases.

There have been significant advances in the use of fMRI technology, but mostly in research. It also comes with significant ethical questions: if you can "read" someone's brain, what are you allowed to do what that knowledge?

For my flight back to the US this weekend I will read two papers: one by Peter Bandettini pubslished in NeruImagine about the history of fMRI and one from Poldrack and Farah on the state of the art in fMRI and its applications, published in Nature.

"Twenty years of functional MRI: The science and the stories, Peter A. Bandettini, Neuroimage 62, 575–588 (2012)

"Progress and challenges in probing the human brain", Russell A. Poldrack and Martha J. Farah, Nature 526, 371–379 (15 October 2015)