In my role working for an AWS consulting partner, I need to keep up to date with new AWS services, features and capabilities, and knowing when it makes sense to use one over another. With the ever increasing pace of innovation, this is an impossible task. Luckily, this is a prime candidate for GenAI, and I’ve been lucky to have preview access to
Amazon Q for a number of months. In this time, using the feature of Amazon Q focused on AWS and embedded in the management console has become an essential application I use in my daily work, and I’m excited for everyone to get a chance to benefit.
In November 2022 OpenAI released an early demo of ChatGPT which rapidly went viral on social media. At its heart is a Large Language Model (LLM) trained on a vast corpus of data, which equates to most of the internet at a specific point in time. The challenge here is the content the model is trained on is typically unverified. It can consist of any opinion an individual has written down and published.
This is where understanding how tools like ChatGPT use a transformer-based deep learning algorithm to generate responses from an input is important. At a high-level, this architecture extracts meanings from text and understands the relationships between words that enables it to predict the next word. This means these algorithms and models do not understand context. As an analogy, imagine a person with no technical knowledge themselves, but who has memorised most of the internet. They will answer a technical question based on this training data, but without knowing how accurate their answer is or whether it makes sense. This is why outputs of tools like ChatGPT can be inaccurate, untruthful, and even misleading at times.
In my line of work, I need to understand the most optimal approach for carrying out a task on AWS. AWS documentation is structured by individual AWS service. If you want to understand more about Amazon DynamoDB then you can read through the service documentation. However, this does not work when you want to understand the differences between services, when you should use serverless over containers, or use a key-value database over a relational database. This is where Amazon Q excels.
The Amazon Q chatbot is trained on 17 years of AWS documentation and blog posts, which provides you with a much greater degree of confidence around the accuracy of the results. It is an expert on patterns in the AWS Well Architected Framework and best practices. Even better is the fact responses include links to top results, which you can click on to read the full document.
I've been using this on a daily basis to give me a starting perspective. I then click the links to read the underlying AWS documentation that supported the response, and validate the accuracy of the answer in my context. I have found this has improved my previous way of working.
But Amazon Q offers so much more. Over time it will help you to troubleshoot and optimise your workloads. It will work with you to draft, review and publish content from one quick prompt. It will also to help with upgrades and code transformation. I’m looking forward to trying out all the new features of Amazon Q announced in the keynote and available in preview. Check it out for yourself, and let me know what you think