Jonathan, before your time at AWS, you founded two start-ups in Berlin. How does working in a smaller startup differ from your work at Amazon?
In essence, it’s not that different. Both the startup and AWS are about constantly questioning conventional ways of thinking. That means every day is different. We want to stay curious, find new solutions to problems and be brave enough to try new things and fail sometimes, because we only learn from our mistakes. At AWS, we call this “Day 1 Culture”. In both cases, agility and innovative strength play a central role.
Innovation: the topic of generative AI is particularly emblematic of this and is on everyone’s lips. What role does this topic play for your teams?
All Amazon customers have been using artificial intelligence when shopping online for 25 years – even if they didn’t realize it until now. And wherever Amazon’s Alexa voice assistant is integrated, AI is behind it. Our teams in Berlin have been working for years on the development of AWS ML and AI cloud services, such as the Amazon SageMaker service, which can be used to train and develop ML models. These ML and AI services are used by over 100,000 business customers around the world, but have perhaps not been as much in the public eye.
2023 was certainly the big year for AI, however, with the technology becoming a mainstream topic. For many, this also marks a starting point for a “golden age” of AI, as the application of AI has evolved and become more tangible to the general public. This is mainly due to the emergence of generative AI, which is driven by new large-scale language models. These language models are trained with many times more computing power and larger amounts of data.
In your opinion, does generative AI also have great potential for German companies?
Generative AI can be a kind of innovation turbo for every industry and this will have an impact on almost every aspect of our lives – from software development to the early detection of cancer. A recent study by AWS and Strand Partners, for example, shows that AI – for the German economy alone – could contribute to additional value creation of 116 billion euros by 2030. So the potential is huge. That’s why we are also working very specifically on new services that make the potential of AI usable for our customers.
Why is AI so relevant for AWS in Berlin and what exactly are you working on?
Amazon has been investing in the development of AI and ML for decades. Germany in particular plays a central role for Amazon. AI/ML research is firmly rooted at our research and development locations in TĂĽbingen, Aachen and, of course, Berlin. One of our major goals in Berlin is to make generative AI usable for everyone via the cloud and to enable our customers, regardless of their size, to use the large language models of the leading providers with their own data so that they get very specific answers or functionalities that are precisely tailored to their business and problem. In essence, we are concerned with how large volumes of information can be processed very efficiently. The first priority for us is that the data is always securely protected.
We first announced Amazon Q in November, and the service is now available in Germany. As an AI assistant for companies, Amazon Q offers enormous advantages for the automation of time-consuming tasks. As one of AWS’s core AI services, Amazon Q was largely developed in Berlin, in close collaboration with our international teams in New York, Seattle and Amsterdam. Many highly talented data scientists and developers work together in our offices in the capital to develop and refine Amazon Q.
In which areas is Amazon Q used? Do you have any specific examples?
Amazon Q can be thought of as a very competent consultant for a company. Amazon Q currently consists of two main components: one is intended for the implementation of business applications, such as chatbots, and the other component was designed specifically for development teams.
As an AI assistant, Amazon Q can help with various typical standard tasks in a corporate context using simple, natural language prompts, i.e. specific instructions or questions. For example, a marketing manager can ask Amazon Q to create a blog post based on an email or to analyze the success of a campaign in detail. For us in Berlin, its use by developers is particularly exciting because Amazon Q is changing the way they use AI and is a real productivity booster in software development. This is mainly because Amazon Q has a deep understanding of its own code and, as a software development expert, can also answer very complex questions precisely in a wide variety of languages. This offers numerous application possibilities for almost every industry.
What does this look like in practice?
Amazon Q can help with the development of new technical features or with the adaptation of old code to new programming language versions, which would otherwise often require thousands of lines of code to be changed manually. Developers can use Amazon Q to debug, test and optimize code, or receive a description in natural language with matching code that they can adopt or adapt with one click. In a test with Amazon Q, a team of just five Amazon developers was able to update 1,000 applications from Java 8 to Java 17 in just two days. In practical terms, this meant a time saving of several weeks, as this would normally have required dozens of development teams.
How can less tech-savvy people try out your technology quickly in their everyday lives?
We didn’t just focus on applications in an enterprise context. A good example of this is PartyRock, an Amazon Bedrock Playground. With PartyRock, we really make it possible for everyone to discover AI in a playful way. You can create your own apps in seconds, for example an app that automatically generates a playlist for a party based on the music tastes of the guests or a cooking app that suggests recipes based on the ingredients in the pantry.
PartyRock was built by just a handful of people in Berlin and was originally intended as an internal tool. Due to the overwhelming response, we developed the AI tool further for external users. To date, over tens of thousands of apps have already been developed with PartyRock. We also work with renowned universities, including the National University of Singapore, which uses PartyRock as a tool for its students to experiment with prompt engineering.
Jonathan Weiss has been working at Amazon for over ten years. As Managing Director Amazon Development Center Germany, he is responsible for the four research and development centers in Germany with over 2,500 employees. His team works at locations in Germany, the Netherlands and the USA. As part of the global development teams, they drive the development of important AI and ML services for Amazon and Amazon Web Services. These include services such as Amazon Q, Amazon SageMaker and PartyRock, which are used by customers worldwide. Before joining Amazon, Jonathan founded the startup Peritor, which was acquired by Amazon in 2012. He studied software engineering and business administration at the Technical University of Berlin and Edinburgh Napier University.
– – – – – –
Further links
👉 https://aws.amazon.com
Photo: AWS
