I hit a barrier developing using generative AI (genAI) technology on my local machine1. As I started to research on a cloud alternative to genAI development, I eventually came across Amazon Q. There are several implementation of Amazon Q – Business, Developer, QuickSights. For this article, I am referring to Amazon Q – Business. Here are my takes on the AWS service so far about Amazon Q Business.
What is Amazon Q – Business?
Amazon Q – Business2 is a managed service that makes it easier to create AI applications. It handles many of the technical details, so you can focus on building your product. For the duration of the article, I will refer to Amazon Q – Business as “Q”.
Why I Liked Q?
There are three things I enjoy about Q: all-in-one, easy to use, and quick setup. What do I mean by all-in-one? You have the ability to get up and running without having to set up other services. Can you leverage other services? Yes. My primary focus was to identify the level of effort (LOE) of cloud genAI technologies that enable creation of simple genAI application.
Q is easy to use. You don’t need to be a genAI expert to get started. From a local development perspective, I required a vector database, a large language model(s), a programming language to create the instructions that enable connection between the data, model, and application layer. That is not the case for Q.
- – I did not have to create and setup a vector database or an intelligence document store like AWS Kendra. You can create a retriever natively within Q.
- – I did not have to set up AWS EC2 or a web server to host my application.
- – I did not have to set up AWS Bedrock to select and configure the model to use.
- – Managing access can be conducted through the Q admin page.
- – You have access to a variety of connectors to some of the most popular systems to include storage services like AWS S3 or external websites.
Q offers a quick setup process. There are countless videos of Q that illustrate how to get up and running. You don’t need an extensive amount of genAI knowledge to get started. However, to maximize the capabilities of the service, it is beneficial to have an understanding of other storage solutions and genAI terms and concepts.
Who Should Use Q?
Everyone can benefit from genAI. However, that is too vague and generic. Let’s take a closer look based on what I know thus far.
From a large organization perspective, it is a great approach to creating proof-of-concept (POC) applications. More specifically, for teams that do not have people with a cloud or genAI background. Q provides a tangible solution that a consumer can get direct access and exposure. When data is an issue, you can create a retriever from an external website. Using other AWS services like Bedrock, Kendra, and S3 with Q can improve the POC functions and capabilities.
Are you a small or midsize organization and are still on the fence of using genAI technologies? Q can provide an easy entry to start experimentation. There are a plethora of resources to walk you through the process of starting. In addition, the amount of dependencies to get set up is minimal.
Do not confuse fast and easy for complete. Regardless of organization size, you require dedicated time to complete a solution that makes an impact to your employees or consumers. First, determine a use case or execute an ideation exercise. In addition, establishing a team that consists of people in the strategic, operational, and tactical zone. This is only a starting point, there are other activities to consider.
Limitations of Q?
Q is a great service. However, it is not with its drawbacks. Let’s take a look at a few items.
Q leverages a subscription model called Pro and Lite. Fundamentally, a pro subscription offers more functionality. As you create a POC, an admin can assign the subscription to each user. However, if an admin assigns a Lite subscription and updates the user to a Pro, the AWS account will be charged at a Pro rate. If you decide to downgrade the subscription, the rate will be applied for the duration of the month and update in the following month.
However, all users need to update a pro subscription by the end of August 30, 20243. It is upsetting because it is a missed opportunity to enable more users access to genAI technology. Everyone does not need a Pro license. Less is more. Alternatively, enable a Lite subscription for development and test environments. Just an idea.
Q enables the ability to create and host a web application. There are ways to customize the application but it is very limited. As of August 16, you are able to update the app title, subtitle, and instruction message. Overall, applying your own branding is limited. Here is a list of items you can’t change:
- AI chat icon and user icon
- Footer message
- Info icon (lower left corner) and message pop up
- No change or option to apply color or theme
Setting up files and folders is critical. My initial testing consisted of using an S3 with a simple file structure and folder naming convention – Listed below:
└── S3 Storage/
├── employee/
│ ├── active/
│ │ ├── employee 1
│ │ └── employee 2
│ └── departed/
│ ├── employee 3
│ └── employee 4
├── employee handbook
├── leave policy
├── maternity and paternity policy
├── pay schedule 2023
└── pay schedule 2024
Q had issues pulling the latest files based on naming convention. For example:
- When asked questions about the current year pay schedule, it would extract the information from pay schedule 2023. At times, it would pull from pay schedule 2024 but that was not very helpful application behavior.
- It struggled with instructions to omit any information from the “departed” folder. Initially, I assume that you can include in your prompt the source location you want to reference but that was not the case.
It may not be a Q issue but the completion of extra steps that involves configuring document metadata. As of this article, I have yet to do a deep dive into more advanced features or tuning from within Q.
Final Thoughts
Q was released April 2024. I only scratched the surface. I look forward to the growth of the service.
Want to know more? I’m here to help! I love building things with tech that make work easier and more fun. Let’s chat and see how genAI can change the way you work!
